EVALUATION OF ROCKFALL HAZARD TO QIRYAT SHEMONA

POSSIBLE CORRELATION TO EARTHQUAKES

MOR KANARI



This work was submitted as M.Sc. Thesis to the Department of Geophysics and Planetary
Sciences, Tel Aviv University

The study was carried out under the supervision of:
Dr. Shmulik Marco, Department of Geophysics and Planetary Sciences, Tel Aviv University
Dr. Oded Katz, Geological Survey of Israel
Dr. RamWeinberger, Geological Survey of Israel





Report GSI /24/08 Jerusalem, September 2008

T
Abstract

his study estimates rockfall hazard for the town of Qiryat-Shemona, situated within the
Dead Sea Transform fault system, at the foot of the Ramim cliff. The 40-m-thick Ein-El-
Assad Formation limestone outcrops provide the source material for rock blocks. Aerial photos
from 19461951 show boulders of volumes of 1 m
3
to 150 m
3
situated within the now built town
premises. This study examines: (a) what are the properties of the source rock and what is the
triggering mechanism? (b) which are the feasible downhill trajectories of the blocks and where
do blocks stop? (c) what is the rockfall hazard, and what may be recommended as a mitigation
design for Qiryat-Shemona?
To answer these questions hundreds of rock-blocks were mapped on the slopes above Qiryat-
Shemona using both field surveys and aerial photos and their volume and spatial distributions
are analyzed; burial ages of soil samples from beneath large fallen blocks were determined by
OSL; rockfall trajectories were simulated using a commercial program (CRSP v4). Hazard
evaluation maps for Qiryat-Shemona were compiled from the results of rockfall simulations.
Simulated analyses of block velocity and kinetic energy may be used as parameters for the design
of mitigation of rockfall damage for Qiryat-Shemona. Rockfall hazard estimation is derived
from: a. rockfall recurrence time based on OSL age determinations; b. block size probability
derived from block volume distribution.
Results show that the block volume distribution follows an exponential function of the form ax
b

with b value 1.17, in agreement with worldwide rockfall inventories. OSL dating of 8 soil
samples demonstrate clustering around dates that coincide with known earthquakes, historic
and prehistoric. It is concluded that earthquakes of large magnitudes (M
w
7) are the triggering
mechanism of rockfalls, yet apparently the rock-mass has to be weakened by joints and fractures
to facilitate rockfalls. Maps of maximal downhill block travel distances combined with slope
morphological analysis were used to suggest possible trajectories of downhill historical rockfalls.
The simulation program variables are calibrated and later used to simulate possible downhill
rockfall block trajectories towards the town premises. Simulation results are used to compile the
rockfall hazard maps. It is concluded that at the south-westernmost part of town, life and
property are at rockfall hazard in particular areas. Rockfall recurrence interval and probability of
block volumes determined from the volume distribution yield hazard evaluation for the area of
Qiryat-Shemona. OSL age analysis of rockfall events (850 years recurrence time and assuming
that the last rockfall triggered by the 1202 AD earthquake) lead to a 6.5% probability for the next
rockfall to occur within the next 50 years, and a 57% probability within the next 475 years.
Evaluated rockfall hazard probability for 50 years is 0.0440.065, and for 475 years 0.385

0.575, for block sizes or smaller than 10125 m
3
respectively. Simulated results of block velocity
and kinetic energy at specific impact locations on town yield block velocities of 1015 m/s and
kinetic energy of 18,00045,000 kJ (98% confidence) for block volume of 125 m
3
. The
recommendation for environmental friendly rockfall damage mitigation design is forestation of
the slope.


Table of Contents


Chapter 1 Introduction............................................................................................................1
1.1 Preface ...................................................................................................................1
1.1.1 Goals and Research Questions .........................................................................2
1.1.2 General Methods.............................................................................................3
1.2 The Study Area.......................................................................................................3
1.2.1 Geological and Seismic Settings for Rockfall ....................................................3
1.2.1.1 Geographic Setting .................................................................................3
1.2.1.2 Geological Setting...................................................................................4
1.2.1.3 Naftali Mountains South to Margaliot Fault .............................................6
1.2.1.4 Naftali Mountains North to Margaliot Fault ...........................................10
1.2.2 Earthquakes in the Study Area.......................................................................12
1.2.3 Rockfalls in the Study Area............................................................................12
1.3 Rockfall Related Literature Review.....................................................................15
1.3.1 Landslides.....................................................................................................15
1.3.2 Rockfalls .......................................................................................................16
1.3.3 Rockfall Initiation Mechanisms (Triggers) .....................................................17
1.3.4 Rockfall Path Profile......................................................................................19
1.3.5 Rockfall Mechanics........................................................................................20
1.3.5.1 Modes of Motion of Rocks .....................................................................21
1.3.5.2 Freefall of Rocks ...................................................................................22
1.3.5.3 Bouncing and Rolling of Rocks..............................................................22
1.3.5.4 Stopping of Moving Rocks.....................................................................22
1.3.6 Rockfall Volume Statistics..............................................................................23
1.3.7 Comparison of Rockfall Models .....................................................................26
1.3.7.1 Empirical Models ..................................................................................26
1.3.7.2 GIS-Based Models.................................................................................26
1.3.7.3 Process-Based Models ...........................................................................27
1.3.8 Rockfall Simulation Programs .......................................................................29
Chapter 2 Methods.................................................................................................................31
2.1 Block Data Acquisition.........................................................................................31
2.1.1 Outline of the Mapping Area..........................................................................31
2.1.2 Block Mapping ..............................................................................................32
2.1.3 Block Volume Measurements.........................................................................33
2.1.4 Diameter Estimation of Non-Mappable Blocks ...............................................33
2.2 OSL Age Determination .......................................................................................35

2.2.1 What is OSL Dating?......................................................................................35
2.2.6 Field OSL Sample Collection..........................................................................35
2.3 Colorado Rockfall Simulation Program..............................................................36
2.3.1 Rockfall Trajectory Analysis...........................................................................36
2.3.2 Rockfall Behavior and CRSP Input Parameters ...............................................37
2.3.2.1 Slope Geometry....................................................................................37
2.3.2.2 Slope Material Properties......................................................................38
2.3.2.3 Rock Geometry.....................................................................................39
2.3.2.4 Rock Material Properties.......................................................................40
2.3.2.5 Starting Zone and Starting Velocity.......................................................40
2.3.2.6 Analysis Points......................................................................................40
2.3.3 CRSP Program Assumptions ..........................................................................40
2.3.4 CRSP Algorithm for Rockfall Simulation........................................................41
Chapter 3 Results...................................................................................................................43
3.1 Source, Blocks and Triggering Mechanism Properties .......................................43
3.1.1 Source Rock Joint Sets...................................................................................43
3.1.2 Rockfall Origin Locations ..............................................................................45
3.1.3 Block Volume Probability Density Function (PDF) ...........................................46
3.1.4 Block Volume Distribution.............................................................................47
3.1.5 Block Estimated Diameters ............................................................................48
3.1.6 Triggering Mechanism of Rockfalls ................................................................51
3.1.6.1 Type of Failure ......................................................................................51
3.1.6.2 OSL Dating Results ...............................................................................51
3.2 Rockfall Trajectories and Affected Area ..............................................................53
3.2.1 Data Collection for this Study ........................................................................53
3.2.1.1 Feasible Trajectories of Falling Rock-Blocks ...........................................53
3.2.1.2 Slope Profiles........................................................................................54
3.2.1.3 Rockfall Trajectories for Calibration ......................................................55
3.2.1.4 Surface Roughness (S)..........................................................................55
3.2.2 CRSP Calibration for this Study .....................................................................58
3.2.2.1 The Delta Max Distance Index...............................................................59
3.2.2.2 Preliminary Calibration Stage ...............................................................60
3.2.2.3 Sensitivity Analysis to Initial Velocity ....................................................60
3.2.2.4 Rock Shape Sensitivity Analysis.............................................................61
3.2.2.5 Calibration Results................................................................................62
3.2.2.6 Calibration Conclusion..........................................................................63
3.2.3 CRSP Simulation Travel Distances .................................................................64
3.2.4 Rockfall Hazard Area Map.............................................................................64
3.2.4.1 Travel Distance vs. Block Volume ..........................................................66

3.2.4.2 Hazard Map of the Distance Distribution of Block Travel........................67
3.2.4.3 Stop Location and Slope Angle ..............................................................71
3.2.4.4 Minimum Shadow Angle.......................................................................72
3.3 Rockfall Impact Hazard .......................................................................................72
3.3.1 Velocity and Kinetic Energy Analysis ..............................................................72
3.3.2 Town Border Impact Hazard Map ..................................................................75
Chapter 4 Discussion .............................................................................................................77
4.1 Triggering Mechanism of Rockfalls.....................................................................77
4.1.1 Nature of the Rockfall Triggering Mechanism................................................77
4.1.2 OSL Results and Past Earthquakes .................................................................78
4.1.3 Clustering of OSL Ages ..................................................................................80
4.1.4 Statistical Validation of OSL Age Clustering ...................................................82
4.1.5 Rockfall Triggering by Earthquakes ...............................................................84
4.1.6 Other Possible Triggering Mechanisms ..........................................................86
4.1.7 Conclusion Rockfall Triggering ...................................................................86
4.1.8 Suggestion for Further Study.........................................................................87
4.1.9 Seismic Topographic Amplification................................................................88
4.1.10 Magnitude of Triggering Earthquakes............................................................88
4.1.11 Rockfall Recurrence Interval .........................................................................89
4.2 Rockfall Trajectories and Affected Area ..............................................................89
4.2.1 Characteristics of Block Stop Locations ..........................................................89
4.2.1.1 Implications for Rockfall Hazard Mitigation Design...............................91
4.2.1.2 Mis-fit of Small Blocks Simulated Travel Distances ................................91
4.3 Hazard from Future Rockfalls .............................................................................93
4.3.1 Rockfall Hazard Area Map ............................................................................93
4.3.2 Town Border Impact Hazard Map ..................................................................94
4.3.3 Frequency of Stopping Blocks and Analysis Location Map ..............................94
4.3.4 Rockfall Hazard Evaluation for Qiryat-Shemona............................................95
4.3.5 Rockfall Block Size Probability.......................................................................95
4.3.6 Rockfall Recurrence Time Probability ............................................................95
4.3.7 Rockfall Hazard Evaluation for Given Time and Block Size.............................96
4.4 Rockfall Impact Mitigation Design......................................................................97
4.4.1 Possible Design for Mitigation of Rockfall Damage.........................................97
4.4.2 Mitigation by Forests .....................................................................................97
Chapter 5 Conclusion.............................................................................................................99
References ............................................................................................................................101
Appendices...........................................................................................................................107

List of Figures
Figure 1-1 View from the Ramim Cliff east toward Qiryat-Shemona ............... 2
Figure 1-2 Topographic setting of the study area, view northwest ............ 4
Figure 1-3 Location map of Qiryat-Shemona ............................................................................ 5
Figure 1-4 Geological map of the study area ............................................................................. 6
Figure 1-5 Main faults of the Hula Valley, Mt Hermon and Southern Lebanon . 8
Figure 1-6 The stratigraphy of the study area .... 9
Figure 1-7 Geological cross sections showing local and regional structures 11
Figure 1-8 Earthquake historical record of the study area ......... 13
Figure 1-9 Israeli standard building code 413 PGA map ......... 13
Figure 1-10 Aerial photo from 1951 blocks at now built town premises ... 14
Figure 1-11 Total number of landslides and earthquake magnitude .... 16
Figure 1-12 A block and its originating cliff behind .... 17
Figure 1-13 Percentage of slope movements for the Yosemite Valley ....... 18
Figure 1-14 Types of triggering mechanisms for slope movements ...... 18
Figure 1-15 Distances of rockfalls from the epicenters of earthquakes . 19
Figure 1-16 Schematic diagram of a typical rockfall path profile ..... 20
Figure 1-17 Ritchies theoretical considerations of rock trajectory ...... 21
Figure 1-18 General modes of motion of rocks on slopes ...... 22
Figure 1-19 Frequency density of Yosemite rockfalls and rockslides .... 24
Figure 1-20 Frequency density and volume of rockfall ............................................................ 25
Figure 1-21 Fahrbschung and minimum shadow angle of a talus slope ................................. 26
Figure 1-22 Actual vs. projected rockfall path ..... 28
Figure 1-23 Rockfall paths resulting from 3D simulation (Apennines site) . 30
Figure 2-1 Block mapping area ............ 32
Figure 2-2 View east towards Qiryat-Shemona ....... 32
Figure 2-3 Correlating maximal GIS dimension to measured block volume . 34
Figure 2-4 Soil samples collection for OSL dating ... 36
Figure 2-5 Measuring surface roughness ..... 38
Figure 2-6 Rock movement analysis ..... 38
Figure 3-1 Location of scan lines at top of the Ein-El-Assad Formation .................................... 43
Figure 3-2 Schematic map of joint sets of the Ein-El-Assad Formation .... 44
Figure 3-3 Overview of the study area facing west ..... 45
Figure 3-4 Location of traversing faults 46
Figure 3-5 Probability density function (PDF) of mapped blocks ...... 47
Figure 3-6 Cumulative probability for given block diameter ................................. 48

Figure 3-7 Maximal GIS dimension of mapped field block volumes ........ 49
Figure 3-8 Correlating maximal GIS dimension with block diameter ... 49
Figure 3-9 Estimated diameters of non-mappable blocks ............................................. 50
Figure 3-10 Anti-slope dips at the base of the source cliff ......... 51
Figure 3-11 All 25 rockfall trajectory simulation profiles used with CRSP ....... 54
Figure 3-12 Measured surface roughness ........ 55
Figure 3-13 Slope with abundant blocks ..... 57
Figure 3-14 Slope with scarce blocks ....... 57
Figure 3-15 Dependence of Surface roughness on the block diameter .... 58
Figure 3-16 CRSP Calibration profiles ..... 59
Figure 3-17 Measuring observed maximal distance ... 60
Figure 3-18 Sensitivity analysis for initial velocity .. 61
Figure 3-19 Sensitivity analysis for rock shape input ..... 62
Figure 3-20 Simulated vs. Observed maximal travel distances ..... 63
Figure 3-21 Rockfall hazard map of Qiryat-Shemona .... 65
Figure 3-22 Maximal travel distances of simulated block diameters .... 66
Figure 3-23 Cumulative frequency of block stop distances N section .... 69
Figure 3-24 Cumulative frequency of block stop distances S section ..................................... 70
Figure 3-25 Rockfall impact hazard at town border line .... 76
Figure 4-1 Grain supported blocks of various sizes at the talus ........ 78
Figure 4-2 Talus blocks vary in size from small fragments to large blocks ... 78
Figure 4-3 OSL age results and candidate triggering earthquakes ....................................... 80
Figure 4-4 Calculated paleo-rainfall amount during the last 7000 years ..... 82
Figure 4-5 Clustering of OSL ages around historical earthquakes .... 83
Figure 4-6 Correlation of cave seismites with lacustrine seismites ....... 85
Figure 4-7 Qiryat-Shemona OSL ages and suggested rockfall triggers ...... 87
Figure 4-8 Slope angle at 50% and 100% of blocks stop ....... 90
Figure 4-9 Observed blocks vs. CRSP simulated travel distances ....... 93
Figure 4-10 Maximum energy dissipated by different tree species ... 98

List of Tables
Table 1-1 Mean recurrence interval for DST earthquakes of M>6 . 12
Table 1-2 Types of landslides ..... 15
Table 1-3 Characteristics of some computer rockfall simulation programs .. 29
Table 2-1 Suggested tangential coefficient input values .... 39
Table 2-2 Suggested normal coefficient input values ..... 39
Table 3-1 Scan-line joint spacing ...... 44
Table 3-2 Block volume (diameter) distribution at mapping area .... 47
Table 3-3 Lab results for OSL ages of Qiryat-Shemona samples ........ 52
Table 3-4 Simulation block sizes ... 53
Table 3-5 Surface roughness field measurements for different rock radii.. .. 56
Table 3-6 Surface roughness for different block sizes .... 58
Table 3-7 Calibration results for CRSP input parameters ..... 64
Table 3-8 Travel distance overshoot ..... 67
Table 3-9 CRSP simulation travel distances .............. 68
Table 3-10 CRSP simulation stop slope angles and swaths ...... 71
Table 3-11 Shadow angle for talus profiles ...... 72
Table 3-12 Calculated velocity and kinetic energy at stop line ...... 73
Table 3-13 Predicted velocity and kinetic energy at town impact line ...... 74
Table 3-14 Analysis point locations vs. stop line and impact line ...... 74
Table 3-15 Predicted velocity and kinetic energy means .... 74
Table 4-1 Rockfall triggering candidate large scale historical earthquakes ..... 79
Table 4-2 Rockfall triggering candidate events and evidence type ... 86
Table 4-3 Slope angle where 50% and 100% of blocks stop ...... 90
Table 4-4 Rockfall block stop location characterization ........ 91
Table 4-5 Cumulative probabilities of selected block diameters ....... 95
Table 4-6 Rockfall hazard evaluation for 50 and 475 years .......... 96

INTRODUCTION
1
R
Chapter 1
Introduction

ockfalls, down-slope movements of rock blocks, are typical in mountainous areas where
they pose a hazard to humans and property all over the world (Crosta and Agliardi,
2004). Fallen rocks on the slopes range in size from cm-size to gigantic boulders hundreds of
cubic meters in size. Rockfalls can be triggered by a variety of factors including earthquakes,
rainfall, freeze and thaw cycles or progressive chemical weathering of rock and discontinuities
(Crosta and Agliardi, 2004). During rockfall events, which are characterized by high energy and
mobility, rock blocks move at velocities up to tens of meters per second. Because of potentially
short spatial and temporal recurrence interval of rockfalls, they are extremely dangerous for
people or property which encounter on their fall track (Crosta and Agliardi, 2004). Thus,
rockfall hazard assessment should be considered for ensuring safety in mountainous areas.
1.1 Preface
The town of Qiryat-Shemona (Figure 1-1) lies in the northern Hula Valley (Figure 1-2). This
valley is part of a series of extensional basins that formed along the Dead Sea Transform active
fault system (Figure 1-3). The town is located at the foot of the fault-controlled Naftali
mountain ridge (Ramim cliff), which rises to its west. New quarters of the town are being
planned and built below the ridge crawling up the slopes. These slopes are spotted with large,
scattered, cliff-derived boulders, which have apparently traveled there by rockfall and landslide
mechanisms. These could have been the results of either one or two processes: one is slow
protracted weathering of the cliff and the other is catastrophic, such as earthquakes or extreme
climatic events like heavy rainstorms or freeze-thaw cycles. Earthquakes are common in the
study area, situated on the western fault branches of the Hula Valley (Heimann and Ron, 1987).
The natural surroundings of the western part of Qiryat-Shemona are prone to rockfall hazard.
INTRODUCTION
2

Figure 1-1. View from the Ramim Cliff east toward Qiryat-Shemona. The town lies at the foot of the cliff, on
the western border fault branch of the Hula Valley (at the background).
1.1.1 Goals and Research Questions
This study aims to evaluate rockfall hazard for Qiryat-Shemona by identifying the rockfall
triggering conditions, modeling the down-slope rockfall trajectories and estimating the expected
kinetic energy of the blocks when reaching town borders. This site serves an example to an
evaluation of rockfall hazard to a settlement lying at the foot of a cliff.
Derived from above are the following required actions:
- Rockfall hazard evaluation for the Ramim cliff slopes.
- Rockfall hazard evaluation for Qiryat-Shemona.
- Suggested design measures for mitigation of rockfall damage to Qiryat-Shemona.
These tasks are addressed by answering the following questions, categorized by rockfall aspects.
Source and Triggering Mechanism
a. What are the geological and seismic settings of the study area?
b. What are the possible triggers for the rockfalls in the Naftali Mountains?
c. Is there a possible correlation of these rockfalls to earthquakes?
d. When have these rockfalls occurred?
e. What is the recurrence interval of rockfalls?
Rockfall Trajectories
f. What are the feasible trajectories of falling rock blocks down the slope?
g. Where do falling rock-blocks stop? Can it be characterized?
INTRODUCTION
3
Rockfall Hazard
h. What is the area subject to probable hazard from future rockfalls on the slopes under the
Ramim cliff?
i. Where are town premises subject to rockfall impact?
j. What may be the rock block size and its possible velocity at the impact point of buildings
or infrastructure of Qiryat-Shemona?
k. What is the hazard to specific properties of Qiryat-Shemona (buildings, roads etc.)?
Rockfall Impact Mitigation Design
l. What are the possible mitigation design measures for rockfall damage?
m. What might be the recommended mitigation design for Qiryat-Shemona?
1.1.2 General Methods
To answer these questions hundreds of rock-blocks were mapped on the slopes above Qiryat-
Shemona using both field surveys and aerial photos and their volume and spatial distributions
were analyzed; burial ages of soil samples from beneath large fallen blocks were determined by
OSL (Optically Stimulated Luminescence) method in order to date the time of the rockfalls;
rockfall trajectories were simulated using a commercial rockfall simulation program (CRSP v4).
Hazard evaluation maps for Qiryat-Shemona were compiled from the rockfall simulations
results. Simulated analyses of block velocity and kinetic energy along the slope suggest the
design for mitigation of rockfall damage for Qiryat-Shemona. Rockfall hazard was derived from:
a. rockfall recurrence time (from OSL age results); b. block size probability (from block volume
statistics).
1.2 The Study Area
1.2.1 Geological and Seismic Settings for Rockfall
Prior to any discussion of the geological and seismic settings converged at the study area, which
may result in rockfall, one simple fact must be faced rockfalls from the cliff of Ein-El-Assad
Formation have occurred in the past. The physical conditions which had matured to result in
these rockfalls have been found in this work to be as follows.
1.2.1.1 Geographic Setting
The town of Qiryat-Shemona is situated at the base of the Naftali mountain ridge, on the
northern part of Israel (Figure 1-3). While the town lies at the north-western border of the Hula
Valley, the slope rises west to 800 m above the valley and 880 m above sea-level (Sneh and
Weinberger, 2003). The outcrops of Ein-El-Assad Formation, the source rocks for rockfall
boulders, are located about 350 m above town. These outcrops create a steep 40 m cliff and a
INTRODUCTION
4
debris slope below it, which reaches town border (Figure 1-2). This slope is covered with
hundreds of scattered boulders that appear to have been detached from the cliff that rises above
the town and driven there by landslide and/or rockfall mechanisms. The boulders vary in size
from tens of cubic centimeters to tens of cubic meters. Here data collection has been performed
in order to evaluate the rockfall hazard for Qiryat-Shemona. Coordinates of the corners of the
study area are (Israels New Grid): SW corner 25200/78600; NE corner 25380/79050.

Figure 1-2. Topographic setting of the study area, view northwest. Top of the Ein-El-Assad Formation is
marked with a red hazard triangle.
1.2.1.2 Geological Setting
The geological background described hereafter is based mostly on the Metulla quadrangle
1:50,000 scale geological map (Sneh and Weinberger, 2003) partly presented in Figure 1-4.
Qiryat-Shemona resides where the western-border fault of the Hula Valley, a pull apart basin
formed along the Dead Sea Transform since early Pliocene (Freund et al., 1970; Garfunkel,
1981), branches into a few faults towards the north (Figure 1-5). The main fault among the
latter is the Qiryat-Shemona fault, which continues northward where it is called the Yammuneh
fault (trending north-northeast). The Yammuneh crosses Lebanon and Syria and continues up to
Turkey (Kafri, 1991; Sneh and Weinberger, 2003). Another branch, the Roum fault trends north-
northwest in Lebanon. The Hula eastern border fault runs along the eastern margins of the Hula
Valley until it reaches Mt. Hermon, where it curves northeastward along the Sion stream
towards Rashiya. Regional stratigraphy (Figure 1-6) is of mid-Jurassic to Holocene, missing
outcrops of late Eocene to mid-Miocene (Sneh and Weinberger, 2003).
The Metulla quadrangle covers an area divided into 7 geological blocks (Sneh and Weinberger,
2003): 1. Naftali Mountains south to Margaliot fault; 2. Naftali Mountains north to Margaliot
INTRODUCTION
5
fault to Qiryat-Shemona fault and further to the Yammuneh fault; 3. from Qiryat-Shemona to
the western foot of Mt. Hermon; 4. western Hermon to Sion fault; 5. eastern Hermon; 6. the
Golan Heights; 7. the Hula Valley. Detailed hereafter are the relevant Naftali Mountains from
each side of the Margaliot fault (blocks 1 and 2 above).

Figure 1-3. Left: Location map of Qiryat-Shemona. Right: Map of the study area in black rectangle. Ein-El-
Assad Formation (EEA) is marked by blue line; town area in red dashed line.
EEA
INTRODUCTION
6

Figure 1-4. Geological map of the study area. Study area is marked by a black rectangle; faults are marked
by solid black lines; concealed faults in dashed lines. Klei Ein-El-Assad Formation which is the source of
the studied blocks; TLS rockfall debris and talus. Digital map data from Geological Survey of Israel (Sneh
and Weinberger, 2003).
1.2.1.3 Naftali Mountains South to Margaliot Fault
Stratigraphy
The slope rising west 800 m above Qiryat-Shemona outcrops these formations (bottom to top):
Hatira, Nebi Said, Ein-El-Assad, Hidra, Rama, Kamon and Deir Hanna. Outcrops of the following
formations are found further south: Sakhnin, Bina, Menuha, Mishash, Ghareb and Taqiye, see
Figure 1-6.
The Hatira and Nebi Said formations comprise the studied slope, and their boundary isnt
conspicuous so they are mapped as a single unit with a 78 m thickness. The mostly sandstone
Hatira Fm. outcrops at its top 50 m, covered by the alternating sandstone, limestone and marl of
the Nebi Said Fm. At the slope base, outcrops are fully covered by recent deposits. The Ein-El-
Assad Fm. is mostly a 42 m thick biomicrite limestone. It outcrops on large parts of southern
Lebanon, where it is dubbed Muraille de Blanche. The Ein-El-Assad Fm. is conformably overlain
by the Hidra Fm. which due to a change in coastline location is comprised of alternating
biomicrite, clay-limestone, sandstone and marl (Sneh and Weinberger, 2002). The Hidra Fm.
forms the upper boundary of the study area.
INTRODUCTION
7
Alluvial deposits originating from slope movements were described in the vicinity of Qiryat-
Shemona (Sneh and Weinberger, 2002) and at the foot of the Naftali Mountains southbound
toward the Kadesh stream. These are apparently recent, inferring from morphological phase of
scarps. Some landslide taluses were spotted south to Qiryat-Shemona.
Tectonic Structure
The Naftali Mountains are bound in the east by the Hula western border fault. The Cretaceous
units of the riding eastern-block dip slightly west towards the synclinal axis of Wadi Dubah. The
mountain-ridge is traversed by faults trending east-west, with only several tens of meters throw.
Two parallel faults trending north-south 500 and 1000 m west to Hula Valley form two
morphological ledges elevated 70 and 300 m respectively above the valley floor.
Recent tectonics is inferred from the locations of faults and fresh scarps, and rock and debris
flows. Shtober-Zisu (2006) concluded that the slope is stable for at least 1.1 Ma. Hence, since
that time, only small-scale tectonic movements (several meters) have occurred on the border
faults, and the basin drainage level is similar to the current. She suggested that since the end of
the early Pleistocene, stream slopes have maintained their morphology and most activity
included deposition and removal of sediments along slope.
INTRODUCTION
8

Figure 1-5. Main faults of the Hula Valley, Mt Hermon and Southern Lebanon. Red lines outline the
geological cross sections A-B and C-D (see text for details); blue rectangle outlines the study area; gray
arrows refer to locations of specified faults; map modified after Sneh and Weinberger (2003).
A
B
C
D
Roum fault
Yammuneh fault
Hasbaya fault
Rashiya fault
Sariya fault
Shamir fault
Margaliot fault
Q. Shemona fault
Tel Hay fault
INTRODUCTION
9

Figure 1-6. The stratigraphy of the study area (after Sneh and Weinberger, 2003).
INTRODUCTION
10
1.2.1.4 Naftali Mountains North to Margaliot Fault
Stratigraphy
The Kurnub and early Judea groups are detailed above. Sneh and Weinberger (2003) add a few
notes (see Figure 1-6): Rama, Kamon and Deir Hana formations are similar on both sides of the
fault. Several other units outcrop here: Avedat group (Timrat and Bar Kokhva formations), Kefar
Giladi formation, Egel Gravel, Hula group (recent rock units, mainly Quaternary, mostly covered
by alluvium), Tufa unit and Sion Conglomerate.
Tectonic Structure
The main faults dictating the tectonic structure of the region are: a. Qiryat-Shemona fault
branching towards the north to Yammuneh and Roum faults; b. Margaliot fault; c. Tel Hay fault;
d. Shehumit fault and its unclear continuation as Hasbaya fault. The eastern border fault of the
Hula valley continues north at Mt. Hermon as Sion fault (Rashiya fault in Lebanon). These
divide the northern margin of the Hula Valley into several tectonic structures (Figure 1-5):
1. Misgav-Am Litani block: between Margaliot fault and Qiryat-Shemona fault (cont. to
Roum fault).
2. Jabel Niha (Lebanon) block: between Roum and Yammuneh faults.
3. Metulla block: between Qiryat-Shemona fault (cont. to Yammuneh) and Tel Hay fault.
4. El Hiyam block: between Tel Hay and Hasbaya faults.
5. Western Hermon block: between Hasbaya and Rashiya faults
Rock bedding of the first 3 blocks above were found intensively folded and faulted by a number
of authors (Glikson, 1966; Kafri, 1989; Picard, 1952; Ron et al., 1997; Weinberger and Sneh,
2004).
Two geological cross sections perpendicular to the above structures were compiled by Sneh and
Weinberger (2003). They are outlined in Figure 1-5 (A-B and C-D) and detailed in Figure 1-7.
INTRODUCTION
11



























Figure 1-7. Geological cross sections showing local and regional
structures. Left-hand is northern (AB) section; right-hand is
southern (CD) section (after Sneh and Weinberger, 2003). See
main faults map above for outline of cross sections.
INTRODUCTION
12
1.2.2 Earthquakes in the Study Area
Situated along the Dead Sea Transform active fault system, the study area is subject to
earthquakes. More than a hundred earthquakes with 3.5<M
L
<6 (local magnitude) were
recorded in the vicinity of the study area in the past 100 years (Figure 1-8). According to Israeli
standard building code 413 peak ground acceleration (PGA) on rock, which has 10% probability
to be exceeded at least once during 50 years (average recurrence interval of 475 years), for the
Qiryat-Shemona Mountains is 0.26 g (Geophysical Institute of Israel; http://www.gii.co.il). The
PGA map is detailed in Figure 1-9. Seismic topographic amplification effect at the Ramim cliff is
discussed in section 4.1.9.
Earthquake Recurrence Interval
The recurrence interval of large (M>6) earthquakes along the DST was analyzed by Begin
(2005) as detailed in Table 1-1. These intervals are referred to later when discussing recurrence
intervals of rockfall for hazard evaluation.
TABLE 1-1 Mean Recurrence Interval for DST Earthquakes of M>6 (after Begin, 2005)
MW ML Years
6.2 6.2 400
6.5 6.5 800
7.0 6.8 3,000
7.5 7.0 10,000
1.2.3 Rockfalls in the Study Area
There is no historical record of rockfalls in the study area. However, rockfalls from the Ein-El-
Assad Formation are observed in the field, while aerial photos dating 1946 and 1951 show rock
blocks of volumes 1 m
3
to 125 m
3
situated within the now built town premises (Figure 1-10).
Rockfall talus is located on the slope west of Qiryat-Shemona, featuring abundant blocks that
have traveled further beyond the talus margin. These observations are the basis for the
evaluation of the rockfall hazard for Qiryat-Shemona.
INTRODUCTION
13

Figure 1-8. Earthquake historical record of the study area. 115 events of local magnitude (ML) > 3.5 for the
years 1905-2008 occurred within in the area outlined in black. Data and figure from the Geophysical
Institute of Israel (http://www.gii.co.il).

Figure 1-9. Israeli standard building code 413 PGA map (Peak Ground Acceleration) for rock. Qiryat-
Shemona is marked in a yellow star (Geophysical Institute of Israel; http://www.gii.co.il).
Earthquake Magnitude
INTRODUCTION
14

Figure 1-10. Aerial photo from 1951 showing blocks where built town premises of Qiryat-Shemona are
located today. Photo area marked with black rectangle in inset map; blocks located on aerial photo using a
stereoscope marked in blue.

INTRODUCTION
15
1.3 Rockfall Related Literature Review
1.3.1 Landslides
Rockfall is one sub-type of the phenomena related to as landslides. The term landslide includes a
wide variety of earth materials movements, such as slumps, soil slips, mud flows, debris flows,
rockfalls etc. (Jibson, 1996).
Landslides classification is frequently based on the type of material involved (soil or rock) and on
the type of movement: falls, topples, slides, slumps, flows, or spreads (Jibson, 1996). Keefer
(2002) studied 40 worldwide historical landslides induced by earthquakes and found rockfalls to
be the most abundant of all landslide types, along with disrupted soil slides and rockslides
(These landslides comprised about 80% of all reported earthquake induced landslides: Rockfalls
35%; disrupted soil slides 26%; rockslides 20% (Keefer, 1984; Keefer, 2002). Keefer (2002)
also showed that landslides were responsible for significant numbers and proportions of
casualties and levels of economic damage in various earthquakes. At least 90% of the deaths in
the 40 earthquakes studied were caused by rockfalls, rock avalanches and rapid soil flows.
Keefer suggests that the high hazard from rockfalls derives from both their transport of rock
material at high velocities and of their abundance.
Varnes (1978) categorized landslides as detailed below in Table 1-2.

TABLE 1-2 Types of Landslides (revised after Varnes, 1978)


Regarding the landslide trigger, summarizing landslide inventories from 1957 to 1999, Keefer
(2002) shows that even moderate-sized earthquakes can produce large numbers of landslides,
and that the number of landslides increases with earthquake magnitude (Figure 1-11).
To summarize, landslides, many times triggered by earthquakes, pose a major natural hazard.
INTRODUCTION
16

Figure 1-11. Relation between total number of landslides and earthquake magnitude. Data are for
worldwide earthquakes between 1957 and 1999 with comprehensive inventories of landslides. The solid line
is the least-squares linear regression (Keefer, 2002).
1.3.2 Rockfalls
Rockfall is a slope process involving the detachment of rock fragments and their fall and
subsequent bouncing, rolling, sliding, and deposition (Varnes, 1978). In some cases rockfall
refers to a quantitative measure describing small phenomena from block falls of a few m
3
up
to 10,000 m
3
events, while rockslides is used to describe more than 100,000 m
3
and rock
avalanches can reach several million m
3
(Dussauge-Peisser et al., 2002). Distinct evidence of
rockfall are talus slope deposits at the foot of steep cliff faces, while such evidence is less distinct
on slopes covered with vegetation.
Rockfall occurs periodically when one or more blocks fall, bounce and roll down a slope. In case
of a talus slope, a block may move further beyond the talus margin, stopping at some distance
from the slope base (Figure 1-12). These blocks pose the greatest hazard in the vicinity of talus
slopes, and the uncertainty of their behavior is a major difficulty in rockfall hazard assessment
(Evans and Hungr, 1993).
Magnitude, M
#

o
f

l
a
n
d
s
l
i
d
e
s
,

N

INTRODUCTION
17

Figure 1-12. A block and its originating cliff behind. The large block has traveled beyond the talus margin.
Photo from the study area: Ramim Cliff, Qiryat-Shemona.
1.3.3 Rockfall Initiation Mechanisms (Triggers)
Rockfall starts with the detachment of rocks from bedrock slopes (mostly a cliff face). Since all
rock outcrops are subject to various degrees of weathering, fracturing leads to jointing and
promotion of rockfall (Dorren, 2003). Apart from weathering, trigger mechanisms also
determine the occurrence of rockfall. Several of triggering mechanisms have been described in
the literature, such as earthquakes (Kobayashi et al., 1990; Vidrih et al., 2001), rainfall and
freeze-and-thaw cycles (Matsuoka and Sakai, 1999).
In their study in the Yosemite valley regarding triggering mechanisms and depositional rates,
Wieczorek and Jger (1996) reported rockfall caused by different factors such as earthquakes,
rain storms, rapid snow melts, freeze-thaw cycles, root penetration and wedging, or stress relief
following deglaciation. Considering precipitation, the highest frequency of slope movements
occurred during the colder and wetter part of the season from November through April (Figure
1-13) and point out the influence of climatic triggering of rockfall.
As for the distribution of triggering mechanisms of slope movements (including rockfalls) of
the recognized triggering events, rain storms (27%) and rapid snowmelt (8%) triggered more
numerous slope movements than earthquakes (5%); however, earthquakes were responsible for
a greater cumulative volume (12%) than these other two individual triggers (Figure 1-14).
Freeze-thaw conditions accounted for only a very small proportion of volume (0.2%) of deposits
from documented triggers (Wieczorek and Jger, 1996).
INTRODUCTION
18

Figure 1-13. Monthly temperature, precipitation and percentage of slope movements for the Yosemite
Valley. Monthly average temperature (C) in dotted line; monthly average precipitation (mm) in black
(right-hand columns); monthly percentages of slope movements exclude those triggered by earthquakes and
those not attributable to any specific month in black left-hand columns of each pair. (after Wieczorek and
Jger, 1996).


Figure 1-14. Types of triggering mechanisms for slope movements. Frequency of occurrence percentage on
the left-hand side; percentage of proportion of cumulative total volume of all slope-movement deposits on
the right-hand side (after Wieczorek and Jger, 1996).
To examine the effect of earthquake magnitude and distance on triggering slope movements in
Yosemite, Wieczorek and Jger used the procedure of Keefer (1984), who developed curves
representing the upper bound of the maximum distance to different types of landslides in
historic worldwide earthquakes of different magnitudes. The solid curve on the plot of
earthquake magnitude versus distance (Figure 1-15) is Keefers historical upper limit for
disrupted slides or falls based on data from 40 worldwide earthquakes. The Yosemite Valley data
falls in agreement with Keefers upper bound curve.
INTRODUCTION
19

Figure 1-15. Distances of rockfalls from the epicenters of earthquakes. Distance to the farthest rockslides or
falls triggered in Yosemite Valley (dots); distance to the center of Yosemite Valley for earthquakes felt in
Yosemite Valley that did not trigger slides or falls there (circles). Numbers refer to earthquakes listed in a
table presented in Wieczorek and Jger (1996). Solid curve is Keefers (1984) upper bound from 40 historical
worldwide earthquakes that triggered slides or falls (after Wieczorek and Jger, 1996).
In the Yosemite compilation, few reports suggest freeze-thaw as probable triggering events;
However, Wieczorek and Jger (1996) suggest that the freeze-thaw process may weaken sites for
subsequent failure by other triggering events, such as rain storms or earthquakes. They conclude
that when triggering events were identified, more slope movements were triggered by large
winter rain storms, rapid snowmelt, and earthquakes than by freeze-thaw conditions or human
activities. Earthquakes accounted for the greatest cumulative volume of slope-movement
deposits from events. Large regional winter storms, with characteristically high daily rainfall and
storm total, have the potential for triggering abundant slope movements.
To conclude, rockfalls may be induced by numerous triggering mechanisms, most of them of
climatic or seismic nature.
1.3.4 Rockfall Path Profile
According to Evans and Hungr (1993), a rockfall-dominated talus slope exhibits a typical profile
as the one seen in Figure 1-16 to which the following description refers. Finer talus fragments
accumulate below the apex (point A). The lowermost part of the talus deposit contains the
largest fragment sizes, as shown between points B and C. Point C is the base of the talus deposit.
Beyond this point the slope is no longer completely covered by talus fragments. The average
talus angle is
1
. The surface to the right of point C consists of material and landforms predating
the talus deposits. The part covered discontinuously by scattered large boulders that have rolled
or bounced beyond the base of the talus (C-D) is referred as the rockfall shadow. The shadow
INTRODUCTION
20
angle is defined as the angle between the outer margin of the shadow and the apex of the talus
slope (
2
) (Evans and Hungr, 1993).

Figure 1-16. Schematic diagram of a typical rockfall path profile. A-C is the talus slope with mean angle
1. C-D is the rockfall shadow with a shadow angle 2 (A-D). 3 is the substrate angle (after Evans and
Hungr, 1993).
1.3.5 Rockfall Mechanics
Ritchie (1963) presented design criteria describing relationships among variables of cliffs, angle
of slopes, depth of ditches, and width of fallout areas. The mechanics of rockfall from cliffs and
talus slopes were dealt with in order to suggest various solutions involving ditches and fences to
contain the rockfall. Regarding talus slopes, Ritchie (1963) observed that the larger the rock is,
the further it gets down slope away from the parent source, and noted that a falling rock spends
most time in the air and would come to a stop when the slope gets flat enough and the
irregularities of the slope are large enough to lodge the rock itself and diminish its velocity. He
suggested a mechanical approach to describe rockfall trajectory, describing the rocks path as a
series of parabolic paths, influenced by the impact angle which determines the velocities along
and normal to the impact plane (Figure 1-17). Ritchie (1963) suggested that the size and shape
of a rock have little bearing on its falling or rolling characteristics. He concluded that since a
falling rock must obey certain natural laws of mass, energy, velocity, impact and restitution,
while being influenced by friction, gravity and time, it seems only reasonable that the behavior of
falling stones would lie between certain limits, limits that can be used as a basis for design to contain
them. (Ritchie, 1963).
Ritchies mechanical considerations relate to a single rock in its movement (Figure 1-17) which
is not influenced by neighboring rock fragments participating in the rockfall. Rockslides, where
a large slope mass of numerous rock blocks moves collectively, are not the case discussed here.
Such cases rely on dynamic velocity analyses and conceptual models of the center of gravity of
the moving mass and several assumptions regarding its starting end ending points, along with
considerations about the influence of particles (blocks) within the total moving rock-mass and
INTRODUCTION
21
other factors (Erismann and Abele, 2001). The model used in this study predicts the down-slope
movement of individual falling rock blocks.
The following sections detail the main aspects of rockfall mechanics which are introduced into
rockfall models.

Figure 1-17. Ritchies theoretical considerations of rock trajectory. Velocity components related to the
moving blocks path and plane of impact (after Ritchie, 1963).
1.3.5.1 Modes of Motion of Rocks
Once the rock has been detached and starts to move, it descends the slope in different modes of
motion that strongly depend on the mean slope gradient (Figure 1-18). The three most
important modes of motion are freefall through the air, bouncing on the slope surface and rolling
over the slope surface. These modes of motion will be shortly detailed in the following sections.

INTRODUCTION
22


Figure 1-18. General modes of motion of rocks during their descent on slopes. Modes are related to the
mean slope gradients (after Dorren, 2003).
1.3.5.2 Freefall of Rocks
Freefall of rocks occurs on very steep slopes. According to Ritchie (1963) freefall occurs if the
slope gradient below the potential falling rocks exceeds 76 but in different field situations this
value varies. Figure 1-18 indicates that around 70 the motion of the rock gradually transforms
from bouncing to falling.
1.3.5.3 Bouncing and Rolling of Rocks
When the mean slope gradient decreases in the down-slope section, a freefalling rock collides on
the slope surface, which is defined as bouncing. During the first bounce rocks tend to break,
especially heavily fractured rocks (Bozzolo and Pamini, 1986). Whether or not a rock breaks,
about 80% of the kinetic energy gained in the initial fall is lost in that first impact (Dorren,
2003).
If the mean slope gradient is less than about 45, a bouncing rock gradually transforms its
motion to rolling because it gathers rotational momentum. During the transition between
bouncing and rolling, the rock rotates very fast and only the edges with the largest radius
maintain contact with the slope. Thus, the center of gravity moves along an almost straight path,
which is an effective mode of motion with respect to energy loss. In fact, this combination of
rolling and short bounces is one of the most economic displacement mechanisms (Erismann,
1986).
1.3.5.4 Stopping of Moving Rocks
A moving rocks velocity (and therefore stopping of a falling rock) mainly depend on the mean
slope gradient. Falling rocks generally accelerate on steeper slopes and decelerate on flatter
slopes (Dorren, 2003). The velocity also depends on the size of the rock and on the material
INTRODUCTION
23
covering the slope such as soil and vegetation. Small rocks slow down more easily than bigger
rocks. First, because during a rockfall the total kinetic energy of small rocks is lower than that of
bigger rocks; secondly large obstacles can more easily stop small rocks; thirdly, small rocks slow
down more easily in depressions between larger rocks on talus slopes. Evans and Hungr (1993)
point out the fact that has been observed by many authors: the biggest rocks are mostly found
near the base of the talus slope.
The friction force of a moving rock is not only dependent on the rock shape, but also on the
surface characteristics of the slope, which is related to the surface roughness (Chang, 1998;
Dorren, 2003; Jones et al., 2000). Surface roughness can be defined as the variation in height
perpendicular to the slope within a certain slope distance (Jones et al., 2000; Pfeiffer and
Bowen, 1989). See section 3.2.1.4 for detailed properties of surface roughness.
1.3.6 Rockfall Volume Statistics
Rockfall inventories are addressed in frequency-volume distributions, unlike other types of
landslides which are usually given in terms of frequency-area statistics (Malamud et al., 2004).
Cumulative frequency-volume statistics for rockfalls have been presented for the Grenoble area
of the French Alps, the Yosemite Valley, USA and worldwide (Dussauge-Peisser et al., 2002;
Dussauge et al., 2003). These authors correlate their frequency-volume data with power-laws of
the form:
b
V V N

~ ) ( (1-1)
With V the rockfall volume, N(V) the number of events greater than V and b a parameter with a
negative slope ranging between 0.40.2 in different worldwide inventories (Malamud et al.,
2004). The equivalent negative slopes for the non-cumulative frequencyvolume distributions
are 1.4 0.2 (Malamud et al., 2004).
Rockfall frequency density function is defined as the number of rockfalls with volumes between
V
R
and V
R
+ V
R
, divided by the width of that bin V
R
:
R
R
V
N
f
o
o
= (1-2)
with f frequency density as function of rockfall volume V
R
, N
R
The number of rockfalls and V
R

the volume bin width.
Guzzetti et al. (2003) examined rockfalls and rockslides in Yosemite Valley, USA and fit their
data to a non-cumulative frequency density function with a negative power-law exponent (b
value) of about -1.1 (Figure 1-19).
INTRODUCTION
24

Figure 1-19. Frequency density of Yosemite rockfalls (squares) and of rockslides (triangles). Black symbols
represent the entire catalogue (1867-2002); Open symbols the period 19802002, for which the catalogue is
compete and reliable. The gray diamond represents 5 prehistoric rock avalanches. The dashed gray line is a
power law fit with exponent, b= 1.1 (after Guzzetti et al., 2003).
Malamud et al. (2004) reviewed rockfall inventories and presented the frequency density of
rockfalls (f = N
R
/ V
R
) as a function of rockfall volume V
R
for three inventories (two of them are
mentioned above): a. 157 rockfalls and slides triggered by the UmbriaMarche earthquakes of
September and October 1997 in Italy; b. 135 rockfalls and slides in the Yosemite region of
California during the period 19802002 (Guzzetti et al., 2003; Wieczorek et al., 1998); c. 89
historical rockfalls in the Grenoble region of France, estimated to date from the period 1904
1996 (Dussauge et al., 2003; RTM, 1997). Also plotted on the same Figure 1-20 is the power law
frequency-volume distribution
37 . 0 log 07 . 1 ) ( log + + =
R R
V V f (1-3)
with f in km
3
and V
R
in km
3
;r
2
= 0.98. (Malamud et al., 2004).
INTRODUCTION
25

Figure 1-20. Dependence of the frequency density f on volume VR for rockfall and rock-slide inventories,
both on logarithmic axes. The straight line is the power-law regression given by log f = 1.07log VR + 0.37
(after Malamud et al., 2004).
The landslide probability density (p) given by Malamud et al. is
L
L
A
N
N
p
o
o 1
= (1-4)
where N is the total number of landslides in the inventory and N
L
is the number of landslides
with area between A
L
and A
L
+ A
L
. Following that, Katz and Aharonov (2006) converted block
surface area distributions resulted from a sandbox shaking experiment to volume distributions
suggesting that rockfall volume probability distribution (p) follows a power law
o
V
NdV
dN
p ~ = (1-5)
where equals 1.13, close to the slope measured for natural rockfall inventories, for which
equals 1.07 (Malamud et al., 2004).
To conclude, the probability density function for rockfall volume p follows a power law of the
form
R
p aV
o
= (1-6)
where a is a linear factor, V
R
is the rockfall volume and equals about 1.1. This has proven to fit
worldwide rockfall and rockslide inventories, implying a possible global correlation between
rockfall volume and its probability density function, which provides the probability for a given
rockfall volume in a given area.
INTRODUCTION
26
1.3.7 Comparison of Rockfall Models
Most of the existing rockfall models for calculating runout zones of rockfall events can be
categorized in 3 main groups: a. empirical models; b. process-based models; c. GIS-based models
(Dorren, 2003). The basic principles of the empirical and GIS-based models are shortly described
in the following sections, since they are not in use in this study, process-based models will be
detailed more extensively.
1.3.7.1 Empirical Models
Empirical rockfall models are generally based on relationships between topographical factors
and the length of the runout zone of one or more rockfall events. Evans and Hungr (1993)
suggest the Fahrbschung principle after Heim (1932) to predict runout zones of rockfall events.
The Fahrbschung is the angle between a horizontal plane and a line from the top of a rockfall
source scar to the stopping point for any given rockfall (Figure 1-21). An alternative principle
suggested by Evans and Hungr (1993) is the minimum shadow angle after Lied (1977). This is
the angle of a straight line between the highest point of the talus slope and the stopping point of
the longest runout boulder for any given rockfall (Figure 1-21). Comparing the results of several
studies, the minimum shadow angle lies between 22 and 30 (Dorren, 2003).

Figure 1-21. Fahrbschung (F) and minimum shadow angle (M) of a talus slope (after Dorren, 2003).
1.3.7.2 GIS-Based Models
GIS-based models are those either running within a GIS environment or they are raster based
models (a row-oriented representation of images where each pixel holds representative data e.g.
elevation or slope angle) for which input data is provided by GIS analysis. GIS-based rockfall
models usually consist of three procedures: a. identification of the rockfall source areas in the
INTRODUCTION
27
region of interest; b. determination of the rockfall trajectory; c. calculation of the length of the
runout zone (Dorren, 2003).
GIS-based models serve as a basis for computerized rockfall simulation programs, which in some
cases use a three dimensional model to compute rockfall trajectories (Agliardi and Crosta,
2003b; Crosta and Agliardi, 2004; Guzzetti et al., 2002; Guzzetti et al., 2004). Such programs
were not used in this study, due to unavailability and extensive high-resolution input
requirements.
1.3.7.3 Process-Based Models
Process-based models describe or simulate the modes of motion of falling rocks over slope
surfaces (Dorren, 2003). Kirkby and Statham (1975) developed a process-based rockfall model
for transport of rocks over talus slopes, assuming that rocks only slide over a talus slope surface.
Based on that model, Keylock and Domaas (1999) developed the simple dynamics rockfall model,
which calculated rock block travel distance over the slope surface on the basis of the friction
force and the acceleration due to gravity. They concluded that the simple dynamics rockfall model
did not appear to hold a significant advantage over the empirical models tested in their study.
Another large group of rather similar process-based models (Azzoni et al., 1995; Bozzolo and
Pamini, 1986; Evans and Hungr, 1993; Kobayashi et al., 1990; Pfeiffer and Bowen, 1989) share
the following: a. models are two-dimensional slope-scale models that restrict falling boulders to
move in a vertical plane (i.e. lateral movements are not simulated); b. the rockfall track is
defined as a series of connected straight lines with a mean slope angle for each such segment of
the rockfall track (Figure 1-22); c. motions are simulated as a series of flying and contact
phases. The flying phase is simulated with a parabola equation based on the initial velocity in x
and y directions and the acceleration due to gravity. The collision point of the rock on the slope
surface is calculated with the intersection of the parabolic flying function and the straight slope
segments. According to Dorren (2003) the major differences between these two dimensional
models are: a. in the representation of the block mass - some represent the falling rock as a
concentrated mass point (Evans and Hungr, 1993; Kobayashi et al., 1990; Pfeiffer and Bowen,
1989), while others consider the block as an ellipsoidal body (Azzoni et al., 1995; Bozzolo and
Pamini, 1986); b. the modes of movement some use detailed characterizations for bouncing,
sliding and rolling (Azzoni et al., 1995; Bozzolo and Pamini, 1986; Evans and Hungr, 1993;
Kobayashi et al., 1990), while others consider bouncing , rolling and sliding as identical
movements that can be described by a series of bounces and impacts (Pfeiffer and Bowen, 1989).
INTRODUCTION
28

Figure 1-22. The upper figure (1) shows the actual rockfall path (a) projected on a height contour line map.
The lower figure (2) shows the slope straight-line segments (b) used in two-dimensional rockfall models
representing the actual slope of the rockfall path (c) (after Dorren, 2003).
Models applying specific algorithms for calculating rolling and sliding velocities mainly use
Coulombs law of friction
| cos - - - = g m F
f f
(1-7)
where F
f
is friction force tangential to the slope surface (kgm s
2
);
f
is coefficient of friction; m
is mass of the rock (kg); g is acceleration due to gravity (9.81) (m s
2
); is mean slope gradient in
degrees (Dorren, 2003).
Kobayashi et al. (1990) developed a model, which used different characterization for bouncing
and rolling, based on the coefficient for efficiency of collision. Their results were all within 30%
range of the measured rockfall runout zones and bounce marks. Their main conclusion was that
block shape is important in governing the modes of motion, but variations in topography control
it (Dorren, 2003).
Pfeiffer and Bowen (1989) developed a model using tangential and normal coefficients of
restitution, with a mass concentrated in one point. Velocity before each impact was resolved into
tangential (parallel to slope) and normal (perpendicular to slope) velocities, which were both
recalculated after impact according to the tangential and normal coefficients of restitution. The
tangential coefficient was determined by vegetation cover and surface roughness, while the
normal coefficient was determined by the elasticity of the surface material. Rolling was
simulated as a series of impact and bouncing events (Dorren, 2003).
All the above-described process-based slope-scale models did not simulate multiple falling rocks
and the complex interactions between them.
INTRODUCTION
29
According to Dorren (2003), process-based models produce more accurate predictions of runout
zones, and seem to be most suitable for application in areas other than the areas were they were
developed and calibrated.
1.3.8 Rockfall Simulation Programs
Several computer programs have been developed and tested to simulate the fall of a boulder
down a slope and to calculate rockfall trajectories (Table 1-3). Most of the programs implement
either a lumped mass (a mass concentrated to a dimensionless point) or a rigid body approach.
For the latter, a simple geometry for the boulder (e.g., a sphere, a disk, a cylinder) is assigned.
Some programs (Azzoni et al., 1995; Bozzolo and Pamini, 1986; Jones et al., 2000; Pfeiffer and
Bowen, 1989) used hybrid approaches, i.e., a lumped mass approach for simulating free fall and
a rigid body approach for simulating rolling, impact and rebound. Most of the programs work
along user-defined slope profiles, and require detailed information on the materials (soil, debris,
rock) cropping out along the slope and the corresponding friction and energy restitution
coefficients. Slope profiles are usually obtained from large-scale topographic maps or through
topographic surveys. Lithological and land-use conditions along the slope are obtained from
field surveys, geological, soil and land-use maps, and the interpretation of aerial photographs
(Guzzetti et al., 2002).
TABLE 1-3 Main Characteristics of Some Computer Rockfall Simulation Programs*
* Table after Guzzetti et al. (Guzzetti et al., 2002). The program used in this study is CRSP v4.
Guzzetti et al. (2002) point out that most of these programs work reasonably well only in small
areas for which detailed thematic information (including topography) is available.
Giani et al., (2004) evaluated two-dimensional and three-dimensional rockfall simulation
models, with CRSP as the 2-D model. They studied Apennines and Alps cliffs (both in Italy) and
pointed out some important conclusions, some of them are brought here:
- Block trajectories calculated by the 3-D model (Figure 1-23) show that the rockfall
phenomenon is characterized by a predominant 2-D development; the rockfall paths
form narrow bundles of mostly parallel trajectories, sub-perpendicular to the cliff
strike. Hence the validity of two-dimensional models is intensified.
INTRODUCTION
30
- The motion efficiency is higher for blocks having spherical shapes and smooth
surfaces than for blocks having irregular surfaces and sections (pointing out a
spherical rock shape as a worst-case scenario, as described later in this study).
- The back analysis performed confirms that both on-site tests and numerical
modeling results are valid instruments to understand this natural phenomenon and
the design of rockfall protection systems.


Figure 1-23. Rockfall paths resulting from the 3D simulation (Apennines site). Paths lead from cliff to
highway. The dimensions and elevation contours are in meters. 3-D trajectory calculations show that the
rockfall phenomenon is characterized by a predominant 2-D development; the rockfall paths form narrow
bundles of mostly parallel trajectories, sub-perpendicular to the cliff strike (after Giani et al., 2004).


METHODS
31
S
Chapter 2
Methods
everal methods of data acquisition and processing used in this study are presented in this
chapter: a. OSL method for determining the ages of rockfalls and helping to elucidate the
triggering mechanism; b. Colorado Rockfall Simulation Program (CRSP v4) for calculating the
trajectories of rockfalls and assessing their impact hazard; c. acquisition of block data from field
and aerial photos used to calculate block size probabilities for rockfall hazard evaluation and
mitigation design.
2.1 Block Data Acquisition
Collection of block data was preformed in a part of the study area (detailed below), where
blocks were mapped, marked and measured for volume. This data was later used to compile the
block size distribution statistics required for hazard analysis.
2.1.1 Outline of the Mapping Area
Block mapping area defined for field data collection is bounded by a number of topographic-
morphologic entities (Figure 2-1). The total mapping area is 110,000 m
2
(on map). Boundaries
of that area are as follows. West: the lower boundary of the rockfall talus (outlined in yellow);
south: a stream running due from the escarpment edge to a road trending north-south, which
also comprises the eastern boundary; north: a small wadi running from the fault at the cliff and
east to the crossroad at the entry point to the geological park, where it meets the eastern
boundary road. Mapping area boundaries are outlined in purple in Figure 2-1.
At first glance, the talus located up slope just below the escarpment holds a vastly larger quantity
of rock blocks than the selected mapping area, so it seems a better candidate for collection of
block volume distribution. But it is that same reason which disqualifies it as representative its
the blocks which have traveled beyond it, although more scarce, which represent a more feasible
block volume distribution in future rockfalls the lower part is where the blocks that travel the
largest distances from the cliff have come to stop (which are still evident today), thus posing the
greatest evident hazard to town (blocks at the eastern, lower parts of the slope have been
masked by the construction of town). Block volume statistics were therefore based on the lowest
part of the slope where blocks are still evident, as detailed below.
METHODS
32

Figure 2-1. Block mapping area. Large blocks can be spotted. Orange arrow marks the point of view of the
next figure below. Inset: black rectangle is map outline.


Figure 2-2. View over the Talus southern boundary east towards Qiryat-Shemona (point of view marked in
the figure above). The yellow line marks the boundary between the talus (middle and left-hand side) and the
mapping area below it.
2.1.2 Block Mapping
Rock blocks were located (GPS coordinates), measured for size (volume), photographed and
marked for further reference. Practically every block larger than 1.0 m
3
(and, in some cases,
even smaller blocks) was accounted for. A catalog of block data was compiled for analysis,
containing 97 blocks (47 other large blocks were documented within the talus area at an earlier
stage of the study, but they were excluded later from the statistics, being situated off the
METHODS
33
mapping area; some of them were used for the past block volume estimation process described
below). Block volumes at the mapping area vary between 0.1 m
3
and 125.0 m
3
. Statistical
analysis was applied only to the 76 blocks of volumes larger than 1.0 m
3
. The smaller blocks are
of negligible importance in terms of hazard assessment anyway.
2.1.3 Block Volume Measurements
Volumes of 144 blocks (97 in the mapping area + 47 in the talus area) were obtained using a
measuring tape. For each block, dimensions in 3 axes were measured and multiplied to calculate
block volume, which should be taken with a grain of salt since no block is really an orthogonal
box shape. The accuracy of the tape is 1 cm and the error of the dimensions measurements is
about 10cm, which may yield a volume error of 15% (~1.5 m
3
) for small blocks (10 m
3
), 10%
(~5 m
3
) for medium sized blocks (50 m
3
) and 5% (7 m
3
) for the largest blocks (125 m
3
). The
full catalog of field observed blocks is presented in Appendix B.
2.1.4 Diameter Estimation of Non-Mappable Blocks
Since Qiryat-Shemona was constructed, blocks which traveled down-slope beyond the current
town border are not available and are not mappable in the field. However, we have a record of
these blocks on aerial photos from 1946 and 1951. Estimation of the size of these blocks (thus
estimating block dimensions predicted to impact the town premises), was carried out in 3 stages:
1. Find a correlation between field measured volume of blocks (3D) and their maximal
dimension (2D) measured on an orthophoto in GIS.
2. Measure maximal dimensions of non-mappable blocks on the 1946-1951 aerial photos
using GIS.
3. Use the correlation found in stage 1 to assign volumes and estimated diameters for the
blocks found in stage 2 (CRSP input size is diameter).
Stage One: Maximal GIS Dimension and Block Volume
58 blocks identified both on the aerials (1946-1951) and in the field were used to correlate
maximal GIS measured dimension to block volume, having their volumes measured. These blocks
were measured for maximal length dimension using the GIS distance measuring tool (0.1 m
accuracy).
Stage Two: Maximal Dimensions of Past-Existent Blocks
Non-mappable blocks (already masked or hidden under vegetation) were identified on the aerial
photos from 1946-1951 and measured for maximal dimension using the GIS distance measuring
tool.
Stage Three: Estimated Diameters of Masked Blocks
Finding a correlation between measured GIS dimensions and estimated diameter:
a. Set a table of block diameters of 0.57.0 m with 0.5 m intervals.
METHODS
34
b. Calculate block volume (spherical) for each diameter in that table.
c. Calculate maximal dimension in GIS for each volume in (b) using the correlation found
in stage one above.
d. Find a correlation between the maximal GIS dimensions found in (c) and corresponding
block diameter in (a) using a linear regression.
This correlation allows assigning an estimated spherical block diameter for every non-mappable
block on the aerial photos from 19461951 by measuring its maximal dimension using GIS.



Figure 2-3. Correlating maximal GIS dimension to measured block volume. Top: blocks identified both on
aerials and in field (red circles); mapping area perimeter in purple; talus perimeter in yellow; inset of
bottom figures in blue. Bottom: blocks measured in GIS for maximal dimension. Notice the identification of
blocks from 1946 (left) and 2007 (right) marked in red ellipses.
METHODS
35
2.2 OSL Age Determination
2.2.1 What is OSL Dating?
OSL stands for Optically Stimulated Luminescence a branch of luminescence, which refers to the
light emitted by some materials in response to some external stimulus processes (Lian and
Roberts, 2006). In the case of OSL, the stimulus is ionizing radiation by visible light. Occurrence
of luminescence has been known for at least 450 years: first scientific description of the
phenomena is related to Gesner (1555), the first experiments to Boyle (1664), and Aitken
(1998) published relevant summaries of luminescence dating. Luminescence dating has been
shown to work for samples as young as a few years, or as old as several hundreds of thousands of
years, and it gives calendar ages directly. The upper theoretical age limit is thought to be a few
million years (Lian and Roberts, 2006). For further detail of the osl dating method, see Lian and
Roberts (2006).
2.2.2 Field OSL Sample Collection
Dating rockfalls using the soil at the interface beneath the blocks bottom face was performed by
Becker and Davenport (2003) by drilling from the top of a fallen block situated on the slope to
the interface at its bottom and radiocarbon dating the soil sample.
In this study, a different method was used. We excavated a ditch beside the block to reach its
interface with the soil, then manually excavated under the block in order to sample the soil at
the interface below its center. Sampling of soil was performed in the dark under a thick wool
blanket cover in order to prevent sunlight exposure from the soil samples (Figure 2-4 a; b).
Dating the rockfalls was based on these assumptions:
1. All sampled blocks did not move from their post-rockfall location. This is based on two
considerations: a. only very large blocks, weighing hundreds of tons, were sampled; b.
the blocks lay on a slope undisturbed by large-scale human activity, where no heavy
mechanical machinery had been operated.
2. Soil samples from the interface between blocks and their underlain soil had been
exposed at the surface during the time of impact of the block, and have been buried
underneath that block ever since.
3. Dating the above soil interface means dating the last time it had been exposed to
sunlight on the surface, which in this case is equal to the time of block impact.
4. The above date of block impact represents the time of the rockfall occurrence, assuming
blocks travel to their post-rockfall location in a very short time span (tens of seconds).
Whether or not the block has reached its current location not from the source cliff, but
METHODS
36
in a secondary rockfall event is less important, because it anyway means a rockfall event
had occurred at that time, which is the required fact in the first place.
OSL tests on quartz fractions from the soil samples, a slope debris component originated at the
sandstone of the Hatira Formation (comprising the slope along which rockfall occurred) were
performed in collaboration with Naomi Porat at the luminescence dating laboratory, Geological
Survey of Israel.

Figure 2-4. Soil samples collection for OSL dating. a. ditch excavation beside the block using a tractor (left);
b. manual excavation of soil sample from interface beneath the block, performed under a blanket cover to
prevent exposure of the sample to sunlight (right).
2.3 Colorado Rockfall Simulation Program
Colorado Rockfall Simulation Program (CRSP V4) is a computer program which simulates rocks
tumbling down a slope (Jones et al., 2000). The program development and algorithm are
described in chapter 2 of the programs User Manual (Jones et al., 2000).
2.3.1 Rockfall Trajectory Analysis
The down-slope trajectory of a rock-block (or the energy lost by a block as it travels down slope)
is affected by slope geometry, slope material properties, rock geometry and rock material
properties (Agliardi and Crosta, 2003a; Ritchie, 1963). Parameters that quantify these factors
are used to simulate rockfall behavior. The model input parameters are slope profile and its
surface roughness, slope rebound and friction characteristics and block morphology. Model
METHODS
37
output is a statistical distribution of velocity, kinetic energy and bounce height along the down
slope trajectory, including stopping distances of the blocks.
Using CRSP provides information on two major aspects of rockfall hazard evaluation. First, the
results of rockfall simulation along a given slope are used to outline areas subject to rockfall
hazard. Second, the statistical distributions of travel velocities and kinetic energy serve as basis
for both hazard analysis and rockfall mitigation considerations. The program needs to be
calibrated for local study area to fit its unique geologic and morphologic surface conditions.
2.3.2 Rockfall Behavior and CRSP Input Parameters
The different parameters which control rockfall trajectories, along with short descriptions of
their affect on rockfall behavior, are described below following the CRSP manual (Jones et al.,
2000).
2.3.2.1 Slope Geometry
Slope geometry is defined by slope angle, slope length, surface roughness and lateral variability.
Slope angle defines the zones of acceleration and deceleration of the fall; Slope length
determines the distances along which rocks accelerate or decelerate. The slope profile is
introduced into CRSP by creating a 2D profile out of cells that have beginning and end X-Y
coordinates. Each cell is a straight line segment which defines its slope angle (). Cell
boundaries are selected where the analyzed natural slope changes its angle or characteristics.
Each profile cell is assigned the parameters defining slope properties, namely surface roughness,
tangential and normal friction coefficients. Irregularities in the slope surface account for most of
the variability observed among rockfall events originating from a single source location. These
irregularities, referred to as surface roughness (S) alter the angle at which a rock impacts the
slope surface. CRSP models surface irregularities by randomly varying the slope angle at block
impact between limits defined by the rock size and surface roughness. Surface roughness (S) is
defined as the perpendicular variation of the slope within a slope distance equal to the radius of
the rock (Figure 2-5). It is a function of rock diameter and the irregularity of the surface used to
estimate the possible variation of the slope angle relative to a given rock size. An increase in
surface roughness has a larger effect on low angle slopes than on steep ones. As surface
roughness increases, velocity generally decreases and bounce height increases (Jones et al.,
2000). During rockfall simulation, the maximum allowable variation in slope angle (
max
) is
calculated for each cell.
|
.
|

\
|
=

R
S
1
max
tan u
where
max
is the maximal allowable variation in slope angle (deg), S is surface roughness (m)
and R is rock radius (m).
METHODS
38

Figure 2-5. Surface roughness (S) measured as the perpendicular variation from average plunge line (
slope angle) over a distance of a rock radius (R). Maximum slope variation (max) is defined by S and R,
Slope angle varies between 0 and max (after Jones et al., 2000).
2.3.2.2 Slope Material Properties
The mechanics of the impact of the rock with slope surface are simulated using two mechanical
coefficients which affect the rebounding of the block. These are represented by numerical values
referred as normal coefficient of restitution (R
n
) and tangential coefficient of frictional
resistance (R
t
) where the normal direction is perpendicular to the slope surface, and the
tangential direction is parallel to the slope surface (Figure 2-6). When a rock bounces on a
slope, kinetic energy is lost due to inelastic components of the collision and friction. R
n
is a
measure of the degree of elasticity in a collision normal to the slope, and R
t
is a measure of
frictional resistance to movement parallel to the slope.

Figure 2-6. Impact angle () defined as a function of rock trajectory, slope angle (), and slope variation
(). Rock velocity (V) is decomposed into normal (Vn) and tangential (Vt) components. The tangential
coefficient of frictional resistance (Rt) and the normal coefficient of restitution (Rn) act to decrease the
falling rocks velocity (after Jones et al., 2000).
For steep slopes where the number of impacts is low, the effect of the material coefficients will
be negligible. For low angle slopes the effect of coefficients is the largest. Jones et al. (2000)
suggest that changes in the coefficients within reasonable limits for a specific slope will not
METHODS
39
produce significant changes in the results. The effect of slope angle and surface roughness is
much greater than the effect of the coefficients. Table 2-1 Lists the CRSP suggested tangential
coefficient values for different slope materials (Jones et al., 2000).

TABLE 2-1 Suggested Tangential Coefficient Input Values (Jones et al., 2000)
Slope Description Tangential Coefficient (Rt)
Smooth Hard Surfaces and Paving 0.90 1.0
Most Bedrock and Boulder Fields 0.75 0.95
Talus and Firm Soil Slopes 0.65 0.95
Soft Soil Slopes* 0.50 0.80
* Soft soil slopes values were extrapolated from other soil types due to lack of data.
Rockfall simulation output is less sensitive to changes in the tangential coefficient (R
t
) than to
changes in the normal coefficient (R
n
). It also suggests that the changes in the tangential
coefficient R
t
are of more significance for hard or heavily vegetated slopes (Jones et al., 2000).
Another major influence on the loss of kinetic energy tangential to the slope is the velocity
normal to the slope. An increase in velocity normal to the surface results in a greater normal
force during impact. The energy lost during the bounce is determined, among other factors,
from the velocity normal to the slope which is calculated based on the normal coefficient (R
n
).
Table 2-2 lists CRSP suggested normal coefficient values for different slope materials (Jones et al.,
2000). Detailed equations regarding the influence of the material coefficients on the behavior of
rockfall are presented in Jones et al. (2000).

TABLE 2-2 Suggested Normal Coefficient Input Values (Jones et al., 2000)
Slope Description Normal Coefficient (Rn)
Smooth Hard Surfaces and Paving 0.60 1.0
Most Bedrock and Boulder Fields 0.15 0.30
Talus and Firm Soil Slopes 0.12 0.20
Soft Soil Slopes* 0.10 0.20
* Soft soil slopes values were extrapolated from other soil types due to lack of data.
2.3.2.3 Rock Geometry
Rock geometry is defined by rock size and rock shape. Since a larger rock has greater momentum
and is less likely to lodge among irregularities, it will travel farther down a slope than a smaller
rock (Ritchie, 1963). Rock size is thus critical in determining the degree to which surface
roughness will affect rockfall behavior (the larger the rock the larger the surface irregularities
required to affect it). Rock size is therefore another factor in determination of surface roughness.
Rock shape contributes to randomness of rockfall and influences the portions of translational
METHODS
40
and rotational energy through the moment of inertia; in CRSP it is restricted to three basic
geometric shapes: sphere, cylinder or disc. Spherical rock shape represents worst-case since it
holds the most mass for a given radius (Jones et al., 2000).
2.3.2.4 Rock Material Properties
Rock material properties are durability and rock mass. Durability determines whether a rock will
break apart upon impact. Two factors act to reduce the influence of rock durability and rock
mass on a rockfall: a. their consistency minimizes their effect on the variability of the rocks
behavior; b. their variation among rocks is considerably less than among slopes or even within a
given slope. Thus the influence of any variability of rock durability or rock mass is negligible
compared to other parameters in the rockfall model.
Rock density varies with rock type. Hoek and Bray (1981) suggest typical rock densities for hard
sedimentary rocks (limestone, dolomite, sandstone) to be 2340 to 2850 kg/m
3
. In this study, all
blocks (limestone) were given density of 2500 kg/m
3
.
2.3.2.5 Starting Zone and Starting Velocity
The starting zone is the rockfall source zone and is defined as a range in the Y (elevation)
coordinates from its top to its base. Starting velocity is comprised of initial horizontal and
vertical components assigned to the rocks. These components are acted upon by gravitational
acceleration until the rocks trajectory intersects the slope, where it is assigned a resultant
velocity for its first impact with the slope.
2.3.2.6 Analysis Points
CRSP allows setting up to three analysis points along the studied slope profile. For each analysis
point, CRSP provides analyses of the following details:
1. Number of rock-blocks passed through the analysis point.
2. The maximum, average, minimum, and standard deviation of rock velocities.
3. The maximum, average, geometric mean, and standard deviation of rock bounce
heights (not used in this study since the authors of the program state that bounce height
analysis is not very reliable (Jones et al., 2000).
4. The maximum, average, and standard deviation of kinetic energies.
5. Cumulative probabilities of velocity, kinetic energy, and bounce height.
6. Graphs of the distribution of rock velocities and bounce heights.
2.3.3 CRSP Program Assumptions
Using the CRSP model for rockfall, the assumptions and constraints stated by the authors of the
program, are hereby brought as elaborated in the manual (Jones et al., 2000).
METHODS
41
On a natural slope, the parameters will have a wide range of values and would be cumbersome
to analyze as independent variables. CRSP reduces the number of variables by means of the
following simplifying assumptions:
1. The slope profile should follow the most probable rockfall path as established during field
investigations. Therefore, all calculations may be in 2-D.
2. Because the rock type does not change during a rockfall and the range of slope material
properties is much greater than that of rock material properties, coefficients assigned to the
slope material (R
n
and R
t
) can account for both the rock and slope properties.
3. The worst case scenario is generally that of the largest rock that remains intact while
traveling down a slope. Therefore, it is assumed that the rock does not break apart in its fall.
4. Rock size and shape are assumed constant for analysis of rockfall from a given source.
Values assigned to these parameters are determined by field study of the source area and
slope materials.
5. For determination of a rocks volume and inertia, a sphere may be used because it yields a
maximum volume for a given radius, which will tend toward a worst case scenario analysis.
2.3.4 CRSP Algorithm for Rockfall Simulation
Rockfall simulation begins within a selected vertical zone representing the source location by
assigning a rock initial horizontal and vertical velocity components. The velocity components are
acted upon by gravitational acceleration until the rocks trajectory intersects the slope below at
resultant velocity V
1
, which is reduced to two components: velocity normal to the slope surface
V
n1
and velocity tangential to the slope surface V
t1
(Figure 2-6). At the point of impact, the slope
angle () is randomly varied up to the limit of the maximum probable variation in the slope
(
max
). At each impact, the incoming velocity, impact angle, and rotational velocity are used along
with the input parameters to calculate new velocity components and rate of rotation. After each
bounce, CRSP performs an iteration to find the time elapsed until the next bounce. Elapsed time is
calculated from X and Y velocity-components, gravitational acceleration, and the slope profile.
After a new impact position is established, the next bounce is calculated. If the distance the rock
travels between bounces is less than its radius, it is considered to be rolling and is given a new
(X, Y) position equal to a distance of one radius from its previous position. This models a rolling
rock as a series of short bounces (Jones et al., 2000). When the impact resultant velocity is
reduced to zero the block stops.
The output of CRSP is a text file containing the simulation analysis. Data relevant to this study
was extracted as follows: a. rock weight calculated from input rock dimensions; b. for each slope
cell: maximal velocity and kinetic energy; c. for each of the 3 analysis points: cumulative
probability statistics of velocity and kinetic energy; number of rocks passed through the analysis
point; d. travel distances (number of blocks stopped per 10 m slope interval).
RESULTS
43
T
Chapter 3
Results
3.1 Source, Blocks and Triggering Mechanism Properties
his section details the joint sets of the source of rockfalls and the results of the OSL age
determination aimed to answer for the triggering mechanism of rockfalls from the Ein-El-
Assad Formation.
3.1.1 Source Rock Joint Sets
Joint sets were measured in the field (location coordinates: E252325/N789515 ITM) performing
3 scan lines at the top of Ein-El-Assad Formation (Figure 3-1), using a measuring tape and a
BRUNTON compass for joint dip measurements.


Figure 3-1. Location of scan lines at top of the Ein-El-Assad Formation.
The analysis of scan lines yielded 3 joint sets excluding the bedding (which is referred to as J1
and was not accounted for). Joint set 2 (J2): 4 joints over 12.5 m with mean joint spacing of
~3.1 m measured along EW striking scan line. Joint set 3 (J3) 7 joints over 7.3 m with mean
joint spacing of ~1.1 m and 9 joints over 7.2 m with joint spacing of ~ 0.9 m along a NS
striking and NNW-striking scan lines, respectively. Joint set 4 (J4) consists of only one joint over
7 m along 305 striking scan line. Schematic representation of joint set orientations are plotted
in Figure 3-2 and detailed in Table 3-1.

East
West
RESULTS
44
TABLE 3-1 Scan-line Joint Spacing
Joint Set Joints/m Mean Spacing Spacing Range
J1 (bedding)* - - -
J2 0.32 3.1 m 2.3 6.3 m
J3 0.93 1.0 m 0.3 2.9 m
J4 ** ? ? ?
* bedding joint spacing was not measured
** J4: only 1 joint along scan-line no spacing data available


Figure 3-2. Schematic representation (map view) of joint sets of the Ein-El-Assad Formation measured at
top of the outcrop. J1 is bedding and was not measured; J2 has a mean spacing of ~3.1 m, J3 has a mean of
1.0 m and J4 spacing data is unknown.
A rough estimation of the block sizes, which are defined by these joint sets, may be obtained
under these assumptions and constraints: a. J4 has little effect on block size since it has larger
spacing than J2 and J3. Therefore, blocks are constrained by the latter two joint sets and the
bedding J1; b. Because quasi-rectrangle fallen blocks were observed on the slope, J1 (bedding)
may have a spacing range similar to those of J2 and J3, e.g. 0.3 6.3 m. the product J1*J2*J3 is
an estimation of block volume. Smallest values (J1J3) yield 2.3*0.3*0.3=0.2 m
3
, while largest
values (J1J2) yield 6.3*2.9*6.3=115 m
3
. These values are in good agreement with direct field
measurements, having blocks of volumes up to 125 m
3
.
Further analysis of the source joint sets was dismissed due to the assumption that de-facto block
volume distribution observations are already available from the rockfall talus and blocks further
down-slope, where block volumes may be obtained directly from field rockfall evidence rather
than theoretically calculated from potential rock mass jointing.
J3 260
= 0.32.9 m
J4 300
=? m
N
cliff free face
J2 360
=2.36.3 m
rock
mass
RESULTS
45
3.1.2 Rockfall Origin Locations
The large rockfall talus evident in the study area and the mapping area where block statistics
data were collected are located just below a negative bottle shaped space (ghost) in the Ein-El-
Assad outcrop mass (Figure 3-3) which seems like a possible source for the rockfall talus. It is
also in very close proximity to a normal fault traversing the outcrop striking west-southwest (all
fault data after Sneh and Weinberger, 2003). This leads to suggest that where rockfall potential
source mass is traversed by faults it is susceptible to long-term deformation and fracturing,
which contribute to the development of mature rockfall conditions (intense fracturing, less
consolidated rock mass). These locations are more susceptible to triggering of rockfalls. The map
in Figure 3-4 outlines such potential locations for rockfall trajectory origins.


Figure 3-3. Overview of the study area facing west towards the Ein-El-Assad outcrop. The rockfall talus in
front. A normal fault at the rock mass can be observed on the top-right (yellow arrow): the overriding block
on the right (north) is lower than the block to the left (south). A scar (yellow ellipse) in the outcrop, possibly
the origin of rockfall mass, is clearly visible left (south) to the fault location. This location is marked in an
orange arrow in the next figure.
Fault
RESULTS
46

Figure 3-4. Location of traversing faults. Source rock mass is subject to fracturing and deformation where
faults traverse it. These locations (yellow arrows) may be more subject to triggering of rockfall. Note that the
studied talus (yellow) and mapping are (purple) where hundreds of fallen blocks are evident, are located just
below a point at the source outcrop traversed by a fault (orange arrow; see previous figure). Faults detailed
after Sneh and Weinberger (2003).
3.1.3 Block Volume Probability Density Function (PDF)
Block frequencyvolume distributions for the observed blocks in the mapping area were based
on the rockfall frequencyvolume statistics presented in the literature review in section 1.3.6.
Block volume statistics for the mapping area were performed using 76 blocks (out of 97 in the
catalog) larger than 1 m
3
.
For practical reasons, block dimensions were measured in the field as box-shaped (L*H*W), yet
the rockfall simulation program input is spherical blocks (section 2.3.2.3). Block diameter
calculation is based on its volume, thus leaving this potential mis-fit on a geometrical shape level
only, which is addressed later in section 3.2.2.4.
The fitted power-law for the probability density function (PDF; Figure 3-5) of the block volumes
in the mapping area is
1.17
0.4
B
p V

=
where V
B
is the block volume in m
3
, with R
2
=0.72. Note that the power in that regression
conforms to that of global inventories (Figure 1-19; Figure 1-20).
RESULTS
47
Block volume (m
3
)
10
-1
10
0
10
1
10
2
10
3
P
r
o
b
a
b
i
l
i
t
y

d
e
n
s
i
t
y

(
m
-
3
)
10
-4
10
-3
10
-2
10
-1
10
0
p = 0.4V
B
-1.17
R
2
= 0.72

Figure 3-5. Probability density function (PDF) of mapped blocks. N=76 blocks (D>1 m
3
) measured for
volume in the field were given probability density values according to their frequency and volume bin size
(section 2.1.3). Plot axes are logarithmic.
3.1.4 Block Volume Distribution
The frequencies of block volumes (or corresponding diameters) were calculated from a total of
N=76 documented blocks. Cumulative frequencies (i.e. the number of blocks at given volume or
smaller) were used to calculate the cumulative probabilities for each block size (Table 3-2).
Since no data was collected for blocks smaller than 1 m
3
(D=1.3 m) there is neither frequency
nor probability available for blocks of that size or smaller.
TABLE 3-2 Block Volume (Diameter) Distribution at Mapping Area
Volume bin (m
3
) Diameter (m)*
Cumulative
frequency*
Cumulative
Probability**
1 *** 1.3 - -
10 2.7 50 0.658
50 4.6 71 0.934
100 5.8 74 0.974
125 6.2 76 1
* Total 76 mapped blocks of volumes 1125 m
3
; volumes correspond to given sphere diameters.
** Cumulative frequency of given size or smaller divided by N (76)
*** No data collected for blocks smaller than 1 m
3

RESULTS
48
The probability values per block diameter (Table 3-2) were plotted and fitted a regression curve
(Figure 3-6), from which probability for any given block size may be calculated, and vice-versa.
The curve x axis is diameter since the rockfall simulation program (CRSP v4) uses diameter as
block size input. Since no data is available for blocks smaller than D=1.3 m, regression was
performed on the cumulative frequencies of 2.7D6.2 m. According to field data, when
rockfall occurs, the probability for a block of given diameter or smaller is
0.412 ( ) 0.262 p Ln D = +
where D is the given diameter (with R
2
=0.97). This may also be used to determine the maximal
block diameter which may be encountered considering a given probability
exp[( 0.262) / 0.412] D p =
where p is the probability one wishes to prepare for. This correlation would come in handy at a
later stage, when considering probable rockfall hazard. Since larger blocks pose larger hazard
due to higher kinetic energy which increases potential damage, the probabilities for different
block sizes may be required when planning rockfall mitigation solutions.
Block diameter (m)
1 2 3 4 5 6 7
P
r
o
b
a
b
i
l
i
t
y
0.0
.2
.4
.6
.8
1.0
2.7 4.6 5.8 6.2
Field data ( >10 m
3
)
Probability fit curve
95% Confidence band
p = 0.412Ln(D) + 0.262
R
2
= 0.97

Figure 3-6. Cumulative probability for given block diameter. Probability curve is calculated from data of 76
blocks (V>1 m
3
) documented in the mapping area.
3.1.5 Block Estimated Diameters
Following the flow for estimating block diameters (section 2.1.4), maximal GIS dimension (x) of
block volume (V) was fitted the power-law x=1.28*V
0.35
with R
2
=0.92 (Figure 3-7).
RESULTS
49
Estimated spherical rock diameter (D) from maximal GIS dimension (x) was fitted a linear
regression D=0.85*x+0.26 with R
2
=1 (Figure 3-8).
Field block volume (m
3
)
0 20 40 60 80 100 120
M
a
x
i
m
a
l

G
I
S

d
i
m
e
n
s
i
o
n

(
m
)
1
2
3
4
5
6
7
8
D=1.28V
0.35
R
2
=0.92

Figure 3-7. Maximal GIS dimension measured for mapped field block volumes.
Maximal GIS dimension (m)
0 2 4 6 8
E
s
t
i
m
a
t
e
d

d
i
a
m
e
t
e
r

(
m
)
0
1
2
3
4
5
6
7
8
D=0.85x+0.26
R
2
=1

Figure 3-8. Correlating between the maximal GIS dimension of mapped blocks and the corresponding
diameter of a spherical block of the same volume. Thus finding a way to estimate the diameters of non-
mappable blocks on aerial photos using GIS (see above text for detail).
RESULTS
50
Estimated diameters assigned to some 200 blocks (Figure 3-9) were later used to determine
observed maximal travel distances required for the back analysis calibration of CRSP v4. Note
that this flow, although statistically and mathematically well-supported, yields an estimation of
a three-dimensional object (a spherical rock) from two-dimensional data (or even one-
dimensional, considering the measured entity is the maximal distance dimension of a 2-D shape).
This leads to suggest that estimated diameters of past existent blocks should be taken with a
grain of salt. However, it is required to make these assumptions in order to acquire an efficient
local-site calibration of the rockfall simulation model, when aiming for valid rockfall hazard
evaluation.

Figure 3-9. Estimated diameters of non-mappable blocks (located using 1946-51 aerials). Blocks in orange
ellipse used for the correlation between the sizes of field and non-mappable blocks.
RESULTS
51
3.1.6 Triggering Mechanism of Rockfalls
3.1.6.1 Type of Failure
According to field observations, the possible types of failure of the rock-mass are toppling or
failure along tension cracks, since sliding along discontinuities (along bedding dip) is not
probable due to the fact that the bedding dips towards the cliff-mass (i.e. opposite to the slope
direction; see Figure 3-10). Therefore, sliding cannot occur as the rock-mass is considered to
have infinite stability against sliding.

Figure 3-10. Anti-slope dips at the base of the source cliff. Bedding (dotted lines) dips toward the cliff mass,
eliminating block sliding failure due to infinite stability against sliding in the slope direction.
3.1.6.2 OSL Dating Results
Lab results of 13 samples include OSL ages of 10 samples taken from blocksoil interface, one
sample taken below blocksoil interface and two pit samples (not directly under a block), as
detailed in Table 3-3. The laboratory report details the following information: Grain size for all
samples is 74-125 m, except for samples QS-1 and QS-2, for which grain size 88125 m was
used. Water moisture estimated at 155%. The quartz was etched by concentrated HF for 40
minutes. D
e
was obtained using the single aliquot regeneration (SAR) dose protocol, with
preheats of 10 s at 220260C and a cutheat 20 below preheat. No. of discs is the number
from those measured that was used for calculating the D
e
(only a small number of measurements
was not used).
10
4m
slope direction
tension crack
RESULTS
52
TABLE 3-3 Lab Results for OSL Ages of Qiryat-Shemona Samples
Sample
ID
Dept
h (m)
+cosm
(Gy/a)
K
(%)
U
(ppm)
Th
(ppm)
Ext.
(Gy/a)
Ext. b
(Gy/a)
Tot. dose
(Gy/a)
No. of
discs
De
(Gy)
Age
(ka)
QS-1 1.2 945 0.8 1.3 6 7 745 1697102 8/10 121.8 7.00 1.10
QS-2* 1.9 836 0.88 1.9 8.5 10 911 1784100 8/9 9912
55.70
7.60
QS-3 0.7 762 1 2 8.4 10 995 176794 6/9 2.70.2 1.50 0.13
QS-4 0.6 577 1 1.6 6.6 8 910 149577 10/13 1.50.3 0.89 0.16
QS-5 0.5 744 0.76 1.3 4 6 680 142985 10/13 5.70.9 4.00 0.70
QS-6 0.4 554 0.46 1.3 4.9 6 520 108064 12/13 2.30.5 2.20 0.50
QS-
7**pit
0.03 536 0.73 1.5 6 7 726 126969 10/13 0.340.10 0.27 0.08
QS-
8**pit
0.3 536 0.73 1.5 6 7 726 126969 9/13 0.890.17 0.71 0.14
QS-9 0.3 455 0.65 1.6 7.1 8 712 117562 10/13 3.61.4 3.00 1.20
QS-
10***
0.5 569 0.66 1.6 7.3 8 722 129971 10/13 10.62.9 8.20 2.30
QS-11 0.3 536 0.58 1.7 6.3 8 665 120967 10/12 2.70.9 2.20 0.70
QS-12 1.2 905 1.08 1.9 9 10 1041 1256108 10/13 8.01.7 4.10 0.90
QS-13 0.4 486 1.05 1.8 8.6 10 1003 148173 12/13 6.83.4 4.60 2.30
* Excluded from statistics because of age axis scaling and large absolute time uncertainty range
** Sampled from pit (not under a block) for current near-surface soil OSL results
*** Excluded from plot and statistics: sampled lower than block-soil interface (30 cm below QS-9)
OSL ages by Naomi Porat, Luminescence dating lab, Geological Survey of Israel

RESULTS
53
3.2 Rockfall Trajectories and Affected Area
Rockfall simulation results are based on 125 simulation runs along 25 profiles, analyzed by CRSP
v4 (Colorado Rockfall Simulation Program). One simulation run consists of 100 blocks simulated
along the profile. Five simulation runs were held per profile one per given diameter (1.3, 2.7,
4.6, 5.8 and 6.2 m; see Table 3-4). This section details the results and analysis of the distribution
of travel distances (or stop distances) while relating to different aspects such as slope angle and
block volume, velocity and kinetic energy. All are summarized for a rockfall hazard evaluation.
TABLE 3-4 Simulation Block Sizes
Simulated Diameter (m) Equivalent Volume (m
3
)
1.3 1
2.7 10
4.6 50
5.8 100
6.2 125
3.2.1 Data Collection for this Study
While data collection for most of the input parameters took place in the field, some parameters
do not require field work. Yet these are still subject to considerations constrained by the field
data. Following is a description of the different CRSP input data elements and how they were
obtained in this study.
3.2.1.1 Feasible Trajectories of Falling Rock-Blocks
Trajectories were selected according to simple assumptions detailed in section 3.2.1.3. In
accordance with the conclusion regarding trajectory origin location (section 3.1.2), 10 out of
the 25 rockfall simulation profiles used in CRSP had their origins located at close proximity to
faults traversing the cliff mass (profiles 5, 6, 8, 9, 19, 21, 25, 26, 28 and 30; Figure 3-11). In
order to obtain a good resolution for rockfall hazard evaluation to town premises, the spatial
density of profiles is much higher (30100 m intervals) where profiles traverse town border (the
northern section of the study area) than at the open areas, where profiles traverse the Route 90
road (150500 m intervals).
RESULTS
54

Figure 3-11. All 25 rockfall trajectory simulation profiles used with CRSP. 10 profiles originate where faults
traverse source cliff. Profiles are denser (30100 m intervals) where town premises are traversed by profile.
Faults detailed after Sneh and Weinberger (2003).
3.2.1.2 Slope Profiles
Obtaining slope profile data was performed on GIS. A Field survey to verify the GIS generated
section of one profile included dip measurements using a BRUNTON compass and a measuring
tape in order to get the correct structure of the source escarpment ledge the Ein-El-Assad
Formation. Once validated, the rest of the slope profiles were extracted using a 5 m elevation
contour map on the system.
RESULTS
55
3.2.1.3 Rockfall Trajectories for Calibration
Feasible rockfall trajectories were defined according to the following: field mapped rock-blocks
found on the slope; mapped blocks from aerial photos dating 1946 and 1951, currently in urban
area; surface data from an orthophoto and topographic maps, 3-D analysis of aerial photos using
a stereoscope and 3-D GIS analysis.

A few assumptions were made in the selection of potential rockfall trajectories, in order to
maintain simplicity and obey basic physics:
1. Trajectories are straight lines originating at the source.
2. Trajectories trend generally in the slope dip direction (deviations from slope dip may
apply according to morphological settings which allow it).
3. Trajectories do not traverse stream paths since a falling block will not climb up from a
stream (which is the lowest point in its vicinity) to higher grounds.
4. Trajectories pass through mapped concentrations of fallen rocks.
In order to follow these rules, a set of maps for the study area was prepared in GIS, including
stream paths, topographic maps, an orthophoto, a DTM, rockfall debris as observed in the field or
extracted from historic aerial photos. A 3-D presentation of the GIS data was also used to suggest
possible rockfall trajectories.
3.2.1.4 Surface Roughness (S)
Surface roughness is measured in the field by stretching a measuring tape down the slope and
measuring the perpendicular distance between the tape and the slope. Within each slope
distance of one-rock radius R, the greatest measurement that occurs several times is the surface
roughness S (Jones et al., 2000). The change in surface roughness for different rock size is
illustrated in Figure 3-12.


Figure 3-12. Measured surface roughness (S). Rock radius R1 (dashed line) yields surface roughness S1;
Greater rock radius R2 (dashed line) yields greater surface roughness S2. For each rock size, the greatest
measurement Si which occurs several times along the slope is the surface roughness.
S2
S1
R1
R2 Slope Surface
RESULTS
56
Surface roughness was measured for rock diameters between 18 m (the largest, very rare,
diameter observed within the study area is about 7 m). Measurements for rock radii from 0.5
4.0 m are detailed in Table 3-5. A number of measurements (S
i
) were performed using a
measuring tape for each rock radius on a few places along the slope. The maximal value of all
given radius measurements (S
max
) is considered its surface roughness (S).

TABLE 3-5 Surface Roughness Field Measurements for Different Rock Radii
All Values are in Meters (m)
Radius (m) S1 S2 S3 S4 S = S
max

0.5 0.04 0.04 0.04 0.03 0.04
1.0 0.13 0.2 0.07 0.11 0.20
2.0 0.14 0.35 0.4 0.15 0.40
3.0 0.15 0.56 0.38 0.3 0.56
4.0 0.26 0.43 0.58 0.56 0.58
* Each table value is observed with some frequency along the slope over a distance of a few rock radii. For
each rock radius, the maximal value of all measurements (S
max
) is considered the surface roughness (S).
RESULTS
57

Figure 3-13. Slope with abundant blocks.

Figure 3-14. Slope with scarce blocks (talus seen at the far right-hand side).
As surface roughness varies with block radius, determination of S for all simulated block
diameters was performed by fitting a logarithmic regression curve to the surface roughness
values measured in the field (Figure 3-15).
RESULTS
58
Block diameter (m)
0 2 4 6 8
S
u
r
f
a
c
e

r
o
u
g
h
n
e
s
s

S

(
m
)
0.0
.1
.2
.3
.4
.5
.6
.7
y = 0.27Ln(x) + 0.03
R
2
= 0.99

Figure 3-15. Dependence of surface roughness on the block diameter. Logarithmic regression is best-fit for
all surface roughness values measured in the field (D=1-8 m); regression equation used later to determine
surface roughness for all simulated block diameters.
Surface roughness values depending on block size (based on a volume scale and converted to
diameters since CRSP block size input is diameter), are detailed in Table 3-6. These values were
later used for all rockfall simulation runs.
3.2.2 CRSP Calibration for this Study
As a first step in performing the hazard analysis, we used the computer program for a back
analysis of known rockfall observations to estimate the input parameters of the analyzed area. In
this study, calibration of the program was performed along 4 slope profiles (Figure 3-16). Each
calibration run was held simulating 100 blocks.
TABLE 3-6 Surface Roughness for Different Block Sizes
Block Volume (m
3
) Block Diameter (m) Calculated S
1 1.3 0.1
10 2.7 0.3
50 4.6 0.4
100 5.8 0.5
125 6.2 0.5
RESULTS
59

Figure 3-16. CRSP Calibration profiles. Town border in red; source in blue; profiles in green; mapped blocks
from field and aerial photos in yellow rectangles (size proportioned to block diameter).
3.2.2.1 The Delta Max Distance Index
In order to validate the calibration of the model, an index which defines the extent to which the
modeled simulation is similar to the field and aerial photos observations was defined: the
stopping location of the simulated blocks can be compared to the observed one (Figure 3-17).
The index named Delta Max Distance (MD) is the subtraction of the field-observed down-slope
maximal travel distance from the simulated maximal travel distance (in meters)

obs sim
MD MD MD = A
where MD is the delta max distance (m); MD
sim
is the maximal simulated travel distance (where
all 100 simulated blocks come to a stop); MD
obs
is the maximal field observed travel distance
(m).
When MD0 the simulation model repeats the field-observations and the model parameters
may be considered calibrated. MD is negative when the simulation undershoots and positive
RESULTS
60
when the simulation overshoots. The determination of the observed maximal distance (MD
obs
)
was based on field block locations and aerial photos from 1946 and 1951, since some parts of
the slope are on currently built town premises where no rockfalls are evident on site.

Figure 3-17. Measuring observed maximal distance (MDobs). Distance for the relevant rock diameter
measured in GIS (blue arrow). Colored circles on right map indicate observed rockfall blocks of different
diameters; blue line is the rock block source; red dashed line indicates the town border in 2008; talus in
yellow; mapping area in green. Background image is the 1946 aerial photo.
3.2.2.2 Preliminary Calibration Stage
Eighty preliminary simulation runs (each with 100 analyzed rock blocks) along the four profiles
were performed using: a. block diameters of 57 m; surface roughness (S) of 0.50.6 (as
measured from field according to block size); b. tangential coefficient (R
t
) of 0.70.9 (0.05
intervals); normal coefficient (R
n
) of 0.120.30 (0.010.02 intervals, since this coefficient has
greater affect on simulation results). Target MD values were set to 60 m (about 10% of
profile). These simulations resulted in the following coefficient value ranges: R
n
=0.20.25;
R
t
=0.70.8. These values were later revised and refined following the initial velocity sensitivity
analysis (detailed below) and a revision in the technique of slope profile extraction in GIS
(yielding better accuracy of the slope topographic section).
3.2.2.3 Sensitivity Analysis to Initial Velocity
An earthquake induces ground acceleration on the blocks. The PGA (Peak Ground Acceleration
according to the Israeli standard building code 413) value for Qiryat-Shemona Mountains is
0.26 g (section 1.2.2). Earthquakes introduce a range of wave frequencies ~110 Hz (Scholz,
2002). The higher frequencies attenuate on shorter distances from the source. The use of 1 Hz
RESULTS
61
frequency denotes a more distant epicenter, but it yields a higher initial velocity (v = a/f), so it
may be used for velocity sensitivity analysis. Hence, considering 0.3 g horizontal acceleration
and assuming 1 Hz earthquake frequency (for highest velocities), the required initial horizontal
velocity can be calculated
/ v a f =
where a=0.3 g (3 m/s
2
) and f=1 Hz, so we get v = 3 m/s.
Sensitivity analysis was performed for two initial horizontal velocities (V
x
): 0 m/s and 3 m/s
(D=6 m sphere; same profile and input parameters: S=0.5 m; R
t
=0.70; R
n
=0.20). Both yielded
similar results for travel distance: MD=-30 m. This test was repeated twice: a. R
n
=0.12 yielded
MD=-90 m for V
x
=0 m/s and -80 m for V
x
=3 m/s; b. R
n
=0.25 yielded MD=+160 m for V
x
=0
m and +150 m for V
x
=3 m/s (Figure 3-18). An exponential regression curve was fit for MD
depending on R
n
values (V
x
= 0 m/s), which yields MD=0 at R
n
=0.215. Consequently, two
conclusions were drawn: a. initial velocity had no significant effect on travel distance; b. the
revised (best-fit) normal coefficient is R
n
=0.22.
Normal Coefficient R
n
0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.26
D
e
l
t
a

M
a
x
i
m
a
l

D
i
s
t
a
n
c
e

(
A
M
D
)
-100
-50
0
50
100
150
AMD = 0 m
V
x
= 0 m/s
V
x
= 3 m/s
y = -98 + 0.33e
26.65x
R
2
= 1

Figure 3-18. Sensitivity analysis for initial velocity (Vx). MD (m) for three Rn values (0.12; 0.20; 0.25) is
plotted per each of two initial velocities: a. Vx=0 m/s (solid circles); b. Vx=3 m/s (hollow circles); zero line
(MD=0 m) is dashed. All other parameters are equal: D=6 m sphere, same profile, S=0.5 m; Rt=0.70.
Exponential regression curve (Vx=0 m/s) fits MD=0 with Rn=0.215.

3.2.2.4 Rock Shape Sensitivity Analysis
For hazard assessment, spherical rock shape represents the worst-case since it holds the most
mass for a given radius and because it rolls more efficiently (Jones et al., 2000). However, this
RESULTS
62
was not revealed in a sensitivity analysis that was performed for three possible rock shapes of
similar weight: a. D=6 m sphere (282,750 kg); b. 4.5x7 m cylinder (278,325 kg); c. 6x4 m disc
(282,750 kg). All were simulated on the same profile with the same input parameters (S=0.50
m; R
t
=0.7; R
n
=0.22). Maximal distances were: sphere 590 m; cylinder 570 m; disc 570 m. This
20 m difference is about 3% of the travel distance, which is negligible. In any case, the spherical
block shape yields worst-case results (farthest traveling blocks). Velocity profiles are plotted in
Figure 3-19.
Distance (m)
0 100 200 300 400 500 600
V
e
l
o
c
i
t
y

(
m
/
s
)
0
5
10
15
20
25
30
Disc
Cylinder
Sphere

Figure 3-19. Sensitivity analysis for rock shape input. Maximal velocities (reported by CRSP) along each
slope cell for each rock shape: a. D=6 m sphere (282,750 kg, black circles); b. 4.5x7 m cylinder (278,325 kg,
gray triangles); c. 6x4 m disc (282,750 kg, gray hollow circles). All simulated on same profile with S=0.50 m;
Rt=0.7; Rn=0.22. Maximal travel distances: sphere 590 m; cylinder 570 m; disc 570 m.
Velocity profiles of all shapes demonstrate almost identical behavior, where maximal velocity
differences among rock shapes are 2 m/s, and in most slope cells are 0-1 m/s. Except for 3 cells
at the higher part of the slope, the spherical block demonstrated higher velocities (if any) along
all profile.
Thus, it is concluded that considering rock shape, a sphere represents the worst-case in terms of
block travel distance and velocity, if such a difference exists.
3.2.2.5 Calibration Results
Further simulations run along the four calibration profiles were now performed for all block
sizes (D=1.36.2 m), using the best-fit coefficient values: S=0.105 m (according to block
RESULTS
63
diameter); R
t
=0.22; R
n
=0.70. All slope cells were given the same values to maintain model
simplicity. The fit between observation and simulation is plotted in Figure 3-20.
200
400
600
800
1000
200 400 600 800 1000
Observed distance (m)
S
i
m
u
l
a
t
e
d

d
i
s
t
a
n
c
e

(
m
)
1:1 line
1-2 m
2-3 m
3-4 m
4-5 m
5-6 m
6-7 m

Figure 3-20. Simulated (4 calibration profiles) and observed (field and 1946-1951 aerial photos) maximal
travel distances. Block diameters are color coded in legend. Middle and large blocks (D3 m; green, orange,
red) demonstrate similar observed and simulated travel distances (close to 1:1 ratio). Small blocks (D<3 m;
blue, light-blue) demonstrate both 1:1 ratio and significantly longer observed travel distances.
Results can be divided into two behavior patterns: 1. mid-size and large blocks (D3 m; green,
orange and red circles): the observed and simulated results are close to the 1:1 ratio; for large
blocks (D4 m), simulated travel distance is a little longer, which yields a more conservative
result and keeps simulation results on the safe side. 2. small blocks (D<3 m): some blocks are
close to the 1:1 ratio, while others demonstrate significantly longer observed travel distances.
3.2.2.6 Calibration Conclusion
In light of the above discussed, the calibrated CRSP input parameter values were determined as
described in Table 3-7. These calibrated values are considered the local site fine-tuning of CRSP
input parameters required for analysis of rockfall hazard for Qiryat-Shemona.
RESULTS
64

An important conclusion from the initial velocity sensitivity analysis is that ground acceleration
has no significant implication on the travel distance of the rockfall blocks. This may lead to
suggest that the main effect of earthquakes in our case-study might be merely the triggering of
the rockfalls.
3.2.3 CRSP Simulation Travel Distances
CRSP v4 simulations block travel distances, detailed in Table 3-9, were analyzed to compile the
rockfall hazard maps and other hazard properties as detailed below.
3.2.4 Rockfall Hazard Area Map
Rockfall hazard map for Qiryat-Shemona is presented in Figure 3-21. The map details the entire
area subject to rockfall hazard, from source escarpment to the west (Ein-El-Assad Formation) to
the stop-line of 100% of blocks. The hazard map was compiled from maximal travel distance
(100% of blocks stop line) of the largest blocks (d=5.86.2 m, V=100125 m
3
) from 25 rockfall
simulation profiles performed using CRSP v4 (detailed in Table 3-9).
TABLE 3-7 Calibration Results for CRSP Input Parameters
Input Parameter Calibrated Value Ranges
Surface Roughness S (m) 0.1 0.5 m (block size dependant)
Tangential Coecient R
t
0.70
Normal Coecient R
n
0.22
RESULTS
65

Figure 3-21. Rockfall hazard map of Qiryat-Shemona. Source of rockfall (Ein-El-Assad formation) marked
in blue line; area subject to rockfall hazard (from source escarpment to 100% of blocks stop line) dashed in
yellow; town border line in red dashed line; Route 90 in orange solid line. Map compiled from maximal
travel distance (100% of blocks stop line) of 25 rockfall simulation profiles performed using CRSP v4 (detailed
in Table 3-9).
RESULTS
66
3.2.4.1 Travel Distance vs. Block Volume
Block diameters simulated for all profiles are detailed in Table 3-4. Travel distances for all
profile are plotted in Figure 3-22 (sections separated for plot detail density).
Travel distance (m)
300 400 500 600 700 800 900
B
l
o
c
k

d
i
a
m
e
t
e
r

(
m
)
1.3
2.7
4.6
5.8
6.2
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
30

Travel distance (m)
100 200 300 400 500 600
B
l
o
c
k

d
i
a
m
e
t
e
r

(
m
)
1.3
2.7
4.6
5.8
6.2
20
21
22
24
25
26
27
28
29

Figure 3-22. Maximal travel distances (100% of blocks stopped) of simulated block diameters for profiles 5-
19 and 30 at the middle section of the study area (top); profiles 20-29 at the northern and southern sections
of the study area (bottom). Profile numbers detailed in legend. For each profile, blocks of all volumes stop at
similar distances from the source cliff.
Out of 125 simulation runs along the 25 profiles (Figure 3-22), an overshoot in the relation of
travel distance and block diameter was obtained in 16 cases (on 11 profiles: 6, 9, 14, 15, 16, 18,
20, 21, 25, 27 and 6) where smaller blocks (i.e. D=1.35.8 m) have traveled farther than the
largest blocks (D=6.2 m). This was resolved by addressing two parameters for each of these
runs: a. the relative overshoot travel distance (calculated by dividing the overshoot distance by
RESULTS
67
the travel distance, thus measuring percentage of overshoot); b. the number of blocks which
demonstrate this behavior (which equals to percentage since all simulation runs were held using
100 blocks). A summary of overshoot distance and block numbers is detailed in Table 3-8.
Since the overshooting percentage of travel distance is very low (mean 2.7%; SD=1.1%) and the
number of overshooting blocks is very small (mean 1.8%; SD=1.4%), it is concluded that
overshooting of smaller blocks in relation to travel distance of the largest blocks (D=6.2 m) is
insignificant and may be neglected.
TABLE 3-8 Travel Distance Overshoot*
Property Min Max Mean SD
Overshoot (m) 10 30 12 5.4
Overshoot (%) 1.2% 4.5% 2.7% 1.1%
No. of blocks* 1.0% 5.0% 1.8% 1.4%
* 16 cases from 11 profiles where D=1.3-5.8 m travel distance was longer than D=6.2 m.
** No. of blocks overshooting beyond the d=6.2 m travel distance out of 100 simulated blocks.
In light of the above it is concluded that hazard analysis may be performed using large block
sizes (diameters). It is trivial that larger blocks also represent worst case scenarios anyway in
terms of mass and kinetic energy. Block sizes are relevant when considering probabilities for
different block diameters, and these are detailed in section 4.3.5.
3.2.4.2 Hazard Map of the Distance Distribution of Block Travel
As concluded above, our hazard analysis is based on the largest blocks (5.8 m, 6.2 m diameter).
The CRSP analysis report contains a frequency count of blocks (out of the 100 simulated blocks
per run), which stop at each 10 m interval of the simulated profile. These frequencies were
summarized (using a MatLab code) to cumulative frequencies of 50%, 68% (1 standard
deviation), 95% (2 standard deviations) and 100% of all blocks, and the distances they were
obtained at are detailed in Table 3-9. Maximal travel distance is determined from the 10 m
interval where 100% of the blocks stopped. 5.8 m blocks simulated maximal travel distance was
usually equal or longer than that of 6.2 m, so this diameter was used for the above calculations of
cumulative frequency distance. In any case the 100% was taken as the longest travel distance,
whether it was for 5.8 m blocks or 6.2 m blocks. Cumulative frequency travel distances (detailed
in Table 3-9) were plotted as map contours (Figure 3-23 and Figure 3-24), which may serve as
basis for statistical considerations of rockfall hazard, e.g. the 95% contour denotes the line
where 95% of large blocks are expected to stop (meaning there is 5% probability for rockfall
impact further beyond this line). Eight simulation profiles (814 and 16, marked by the sword
in Table 3-9) at the south-western part of town demonstrate an intersection of maximal travel
distance with town border line, which means that town premises at these profile trajectories are
subject to rockfall impact hazard. Distances at which blocks impact town border line are also
RESULTS
68
detailed in Table 3-9 to compare against distances covered by the cumulative frequencies of
blocks and their travel distances.

TABLE 3-9 CRSP Simulation Travel Distances
Profile* 50% stop (m)** 68% stop (m)** 95% stop (m)** 100% stop (m)**
22 280 280 290 300
21 380 390 400 420
20 300 310 320 350
19 350 350 360 360
30 370 380 390 400
5 390 390 400 440
18 420 430 450 460
17 430 440 450 460
15 490 490 500 520
16 (590 m) 560 570 580 590
14 (510 m) 530 530 540 550
13 (490 m) 570 570 590 590
12 (490 m) 570 570 590 590
11 (500 m) 580 580 590 600
10 (520 m) 540 550 560 560
7 500 510 520 600
6 460 650 680 710
8 (780 m) 830 840 850 850
9 (780 m) 810 830 840 850
24 150 500 550 570
25 10 180 210 220
26 410 420 430 450
27 330 340 360 370
28 240 250 260 270
29 50 180 190 210
* Profiles are ordered according to their map location from north (top of table) to south;
** Distance at which stated percentage (median, +1SD, +2SD, 100%) of simulated blocks stop; 100% data
from simulating 5.8 m and 6.2 m diameter (longest travel distance per profile); all other data from simulating
5.8 m diameter (100 m
3
) blocks.
Simulated travel distance intersects town border line, introducing rockfall impact hazard to town premises.
Travel distances in brackets are where blocks impact town border line.

RESULTS
69

Figure 3-23. Cumulative frequency of block stop distances northern section of study area. Contours
denote where given percentage of large falling blocks is expected to stop. Lines are smoothed interpolations
from all simulation profiles using 5.8 m block diameter (except 100% line which is comprised of the longest
travel distance among 5.8 m or 6.2 m). Orange-hazard triangles are locations of maximal travel distances
(100% of blocks) and are labeled per profile; Route 90 marked in dashed red line only outside town
premises; see legend for other details.
RESULTS
70

Figure 3-24. Cumulative frequency of block stop distances southern section of study area. Contours
denote where given percentage of large falling blocks is expected to stop. Lines are smoothed interpolations
from all simulation profiles using 5.8 m block diameter (except 100% line which is comprised of the longest
travel distance among 5.8 m or 6.2 m). Orange-hazard triangles are locations of maximal travel distances
(100% of blocks) and are labeled per profile; see legend for other detail.
RESULTS
71
3.2.4.3 Stop Location and Slope Angle
Data regarding the angles of slope cells where all blocks stop (100% stop angle) and the distance
blocks travel within this cell till all come to a stop (stop swath) were extracted from CRSP
simulation analysis reports (Table 3-10).
TABLE 3-10 CRSP Simulation Stop Slope Angles and Swaths*
Profile Stop Slope Angle (deg.) Stop Swath Distance (m)
5 10 13
6 10 87
7 9 33
8 6 42
9 5 49
10 4 8
11 8 40
12 5 11
13 5 11
14 7 38
15 9 23
16 9 51
17 8 22
18 10 48
19 3 65
20 12 34
21 9 41
22 7 36
24 8 10
25 7 28
26 11 105
27 8 14
28 7 61
29 7 41
30 8 33
* Stop angle refers to simulation cell where 100% of blocks stop; stop swath refers to distance within that
cell at which 100% of blocks stop. Mean stop angle is 7.7 (SD=2.3); mean stop swath is 38 m (SD=24 m).
RESULTS
72
3.2.4.4 Minimum Shadow Angle
Minimum shadow angle is defined by the angle of a straight line between the highest point of a
talus slope (just below the source cliff base) and the stopping point of the longest runout
boulder. Typical minimum shadow angles lie between 2230 (see section 1.3.4). Minimum
shadow angles calculated for CRSP simulations and for observed blocks from field and aerial
photos (only for profiles traversing the talus located in the study area) are detailed in Table 3-
11. The mean simulated shadow angle is 21.5 (SD=1.2) and the mean observed shadow angle
is 18.0 (SD=1.7).
TABLE 3-11 Shadow Angle for Talus Profiles*
Profile Simulated Shadow Angle Observed Shadow Angle
5 20 16
15 22 17
16 22 17
17 22 20
18 23 20
30 20 18
*calculated minimum shadow angles for CRSP simulation results stop line and for farthest traveling blocks
observed from 1946-1951 aerial photos (only for profiles traversing talus). Mean simulated shadow angle is
21.5 (SD=1.2); mean observed from aerials is 18 (SD=1.7)

3.3 Rockfall Impact Hazard
3.3.1 Velocity and Kinetic Energy Analysis
The CRSP analysis report details block velocity and kinetic energy, which can be used to plan
defensive measures and design mitigation facilities against rockfall damage. Kinetic analyses
provided by CRSP with 98% cumulative probability (i.e. the maximal velocity expected with 98%
confidence level) were used in order to obtain worst-case hazard evaluation.
It is pointless to evaluate these kinetic properties at stop line, because the velocities at stop
distances are null. Hence, an evaluation of the kinetic properties was performed by locating CRSP
analysis points at 95% of the maximal travel distance (within 10 m distance or less from 95%
travel point, since CRSP output distance resolution is 10 m). Where simulated blocks impact town
border before stopping (profiles 814; stop line of profile 16 is located just at town border), stop
line is not valid for analysis since there is no way to simulate the movement of blocks through
town property (buildings, paving etc.) until they reach the theoretical simulated stop line.
Hence, kinetic energy analysis for these profiles was performed at town border, while applying
the same considerations stated above regarding 95% of maximal travel distance (considering
inner coherence within kinetic analysis results). Predicted block velocity and kinetic energy
RESULTS
73
(largest block, d=6.2 m; 98% cumulative probability) are detailed in Table 3-12 and Table 3-13.
5% distances from stop line and analysis point distances from stop line are also detailed. Velocity
and kinetic energy analyses at stop line or impact line are detailed per profile in the engineering
appendix (Appendix A).
TABLE 3-12 Calculated Velocity and Kinetic Energy at Stop Line*
Profile 5% distance (m)** Analysis pt dist. (m)** Velocity (m/s) Kin. Energy (kJ)
22 15.0 20 5.1 4,423
21 20.5 20 6.2 6,541
20 17.0 20 4.6 4,479
19 18.0 20 7.3 8,740
30 20.0 20 5.7 5,464
5 22.0 20 5.6 6,632
18 22.5 20 7.0 8,086
17 23.0 30 6.7 9,180
15 25.5 20 6.3 6,576
16 29.0 20 6.7 7,445
7 30.0 30 4.5 4,334
6 34.0 30 8.9 12,737
24 27.5 20 8.6 11,972
25 10.5 10 6.7 7,598
26 22.5 30 6.2 6,492
27 18.0 20 6.8 7,619
28 13.5 20 5.7 5,400
29 10.5 20 4.9 4,196
* Predicted velocity and kinetic energy (98% confidence level) for profiles with no impact at town border.
** Distance between 95% travel distance and stop line; CRSP analysis point located within 10 m distance or
less from 95% travel point. See text for detail.
Mean values of 5% distances from stop line compared against analysis point distances from stop
line (or impact line) are detailed in Table 3-14.
Mean velocity and kinetic energy predictions at the stop line and town border impact points
(depending on profile) are detailed in Table 3-15 and in the engineering appendix (Appendix
A). At stop line, velocity varies between 4.59 m/s and kinetic energy between 4,20012,800 kJ;
at town impact points velocity varies between 1015 m/s and kinetic energy between 18,000
45,000 kJ (design of energy dissipation systems is discussed in section 4.4).
RESULTS
74
TABLE 3-13 Predicted Velocity and Kinetic Energy at Town Impact Line*
Profile 5% distance (m)** Analysis pt dist. (m)** Velocity (m/s) Kin. Energy (kJ)
14 39 40 11.9 28,762
13 39 40 14.8 44,479
12 26 30 9.7 17,944
11 25 30 13.9 40,172
10 24.5 30 11.9 29,576
8 24.5 30 12.2 30,441
9 25.5 30 11.1 24,821
* Predicted velocity and kinetic energy (with 98% confidence level) for profiles at impact with town border.
** Distance between 95% travel distance and stop line; CRSP analysis point within 10 m distance or less from
95% travel point. See text for detail.
TABLE 3-14 Analysis Point Locations vs Stop Line and Impact Line*
Property Mean (m) SD (m)
Stop line 5% of travel deviation 21.1 6.6
Analysis point deviation 21.7 5.1
Impact line 5% of travel deviation 29.1 6.8
Analysis point deviation 32.9 4.9
* Analysis points located within 10 m distance from 95% of travel distance due to CRSP distance resolution.
5% of travel deviation is the distance between maximal travel distance and 95% of it. Analysis point
deviation is the distance between analysis point and maximal travel distance (up to 10 m from the 5% travel
deviation).
TABLE 3-15 Predicted Velocity and Kinetic Energy Means
Property Mean SD
Stop line Velocity (m/s) 6.3 1.2
Kinetic Energy (kJ) 7106 2437
Impact line Velocity (m/s) 12.2 1.7
Kinetic Energy (kJ) 30,885 8960
* Predicted velocity and kinetic energy (with 98% confidence level) for profiles at impact with town border.
The above values may be used as general guidelines for rockfall damage mitigation for the entire
study area, yet for specific locations it should be based on predicted velocity and kinetic energy
at specific locations as in Table 3-13.
RESULTS
75
3.3.2 Town Border Impact Hazard Map
CRSP stop line (100% of blocks stop) and CRSP velocity and kinetic energy analysis points at town
border (at south-western part of town, where simulations yield rockfall impact at town area) are
plotted in Figure 3-25. For kinetic analysis detail (block velocity and kinetic energy), locations
marked by profile indexes refer to Table 3-12 and Table 3-13, and the engineering appendix
tables.
RESULTS
76

Figure 3-25. Rockfall impact hazard at town border line. CRSP stop line in yellow, constructed of a smoothed
line of stop locations of largest blocks (5.86.2 m diameter) of all profiles; town border in red dashed line;
orange-hazard triangles mark simulated stop line for each profile; profile indexes in yellow refer to profile
numbers in the engineering appendix tables; indexes with sword label mark locations of rockfall impact
at town border; inset: black rectangle outlines map area.

DISCUSSION
77
T
Chapter 4
Discussion
4.1 Triggering Mechanism of Rockfalls

his section discusses issues related to the source of the rockfall blocks and the triggering
mechanism of rockfalls from the Ein-El-Assad Formation.
4.1.1 Nature of the Rockfall Triggering Mechanism
Most rockfalls are associated with triggering events, such as earthquakes, rainstorms, or periods
of warming producing a rapid melting of snow. The magnitude and proximity of the earthquake,
intensity and duration of the rainfall, thickness of the snow-pack, and the warming pattern, all
influence the triggering of rockfalls (Wieczorek and Jger, 1996). However, some rockfalls occur
without a direct correlation to an obvious triggering event; Wieczorek et al. (1995) suggest that
such rockfalls occur in granite probably due to processes associated with gradual stress release
and exfoliation.
Evidence for a catastrophic (sudden, as opposed to natural rock weathering resulting in toppling
of blocks) nature of rockfalls is observed all over the talus area (Figure 4-1 and Figure 4-2):
Blocks of varying volumes, from the smallest pebbles to boulders of tens of cubic meters in
volume, are grain-supported (i.e. leaning in direct contact against each other) and in many cases
show impact deformations on their common faces. This indicates that these aggregations of
blocks have traveled to their current location simultaneously. The presence of all block sizes also
supports that these are outcrops of large-scale rock mass movements, which include fracturing
and disintegration of the blocks into smaller fragments. A slow, natural erosion mechanism
could not have resulted in such aggregations, but in large, matrix-supported blocks (i.e.
detached from each other by slope soil or debris), with little or no evidence for fragmentation
into the smallest fragments.
The above conclusion supports the possibility that the rockfalls were triggered by discrete events
such as earthquakes or extreme precipitation events.
DISCUSSION
78

Figure 4-1. Grain supported blocks of various sizes at the talus. Marked length is approx. 70 cm.

Figure 4-2. Talus blocks vary in size from small fragments to large blocks. Yellow stick is 40 cm.
4.1.2 OSL Results and Past Earthquakes
Rockfall triggering candidates of strong earthquakes for the past 3,000 years were listed using
data from earthquake catalogs (Amiran et al., 1994; Guidoboni and Comastri, 2005; Guidoboni
et al., 1994). The criteria for catalog earthquakes to qualify as rockfall triggering candidates are:
a. their estimated maximum intensity is at least IX on an EMS macroseismic local intensity scale
(Amiran et al., 1994; Guidoboni and Comastri, 2005; Guidoboni et al., 1994); b. their calculated
moment-magnitude is 6.5 or larger and the distance between the study area and affected
70 cm
40 cm
DISCUSSION
79
localities reported in the catalogs does not exceed 100 km. This condition satisfies Keefers
(1984) upper limit for disrupted slides or falls triggered by earthquakes based on data from 40
worldwide earthquakes (section 1.3.3). Earthquakes are detailed in Table 4-1. Two large
earthquakes, 1759 AD and 1837 AD, are excluded from the table since no block was dated to later
than 1300 AD (including uncertainty range).
TABLE 4-1 Rockfall Triggering Candidate Large Scale Historical Earthquakes*
Date Age (ka) Max Local Int. Affected Localities
759 BC 2.77 - Israel**
199 BC 2.21 X Sidon
363 AD 1.64 IX Banias, Caperaum, Tiberias, Hamat-Gader
502 AD 1.51 X Beirut, Sidon, Ptolemais (Akko)
551 AD 1.46 VIII-IX Sidon, Beirut, Jerusalem, Petra, Jarash
659 AD 1.35 IX Palestine, Syria
749 AD 1.26 X
Capernaum, Susita, Bet She'an, Tiberias, Hamat Gader, Lod,
Jerusalem
1033 AD 0.97 IX-X Syria, Palestine, Tiberias, Galilee, Jericho
1202 AD 0.81 IX
Safed, Tiberias, Banias, Baal-bek (Lebanon), Bet Shean,
Akko, Bet Jan
* Earthquake data from historical earthquake catalogs (Amiran et al., 1994; Guidoboni and Comastri, 2005;
Guidoboni et al., 1994)
** Earthquake cited in the book of Amos; magnitude estimated to 7.3 by Migowski et al (2004)
Migowski et al. (2004) established a high resolution Holocene seismic history of the Dead Sea
Transform based on intervals between deformed laminated sediments (seismites). The intervals
between seismites were determined by varve-counting in laminated sedimentary cores and
radiocarbon dating of seismites, thus correlating deformed sequences in Ein-Gedi (Dead Sea) to
earthquakes and their magnitudes. Their profile of the Holocene indicates recurrence intervals
of seismic activity varying between a few and 1,000 years, with a conspicuous minimum rate at
210031 BC and a noticeable maximum during the past six to eight centuries. They suggest that
most of the epicenters of the correlated earthquakes are situated lose to the Dead Sea (within
150 km) or up to 400 km north of it along the DST. They suggest an earthquake of magnitude 7.3
in 759 BC and two earthquakes of magnitude 8 (with uncertainty) and 6.8 in 2050 BC and 2100
BC respectively (Migowski et al., 2004) calculated from local intensities according to former
studies (Amiran et al., 1994; Salamon et al., 2003). These two suggested events fit OSL ages of
soil samples found in the current study, detailed later.
The data are summarized in Figure 4-3, which describes OSL ages for the sampled blocks of
Qiryat-Shemona for the past 10,000 years.
DISCUSSION
80
Age (ka)
0 1 2 3 4 5 6 7 8 9 10
S
a
m
p
l
e

I
D
QS-1
QS-3
QS-4
QS-5
QS-6
QS-9
QS-11
QS-12
QS-13
Yagoda et al
Kagan et al
1
-2000 -1000 0 AD +1000 +2000

Figure 4-3. OSL age results for block samples and candidate triggering earthquakes for the past 4,000 years.
Age lab results of 9 sampled blocks plotted as black circles with error bars; ages of earthquakes determined
by Yagoda et al. (2007) in gray diamonds and by Kagan et al. (2005) in gray triangle; dates of candidate
historic large-scale earthquakes in solid gray lines (Table 4.1); candidate earthquake from seismic record in
dashed gray line; Christian calendar time scale at top axis (+ is AD, is BC). Block samples QS-2 and QS-10
are excluded (see detail in text).
Two block samples were excluded from the plot: a. QS-2 which is dated to 55.77.6 ka (excluded
for a very large uncertainty range); b. QS-10 which is dated to 8.22.3 ka, which was sampled at
the same block as QS-9 but 30 cm below blocksoil interface. Also not included in the plot are
QS-7 and QS-8, sampled at a pit for control.
4.1.3 Clustering of OSL Ages
There are three possible interpretations to the OSL results, brought here with supportive pros and
possible cons:
a. Each age represents a different rockfall event triggered by a different earthquake. Pros: a
limited number of blocks were sampled (9 ages on which considerations are based) and
there are several earthquakes serving as candidates for rockfall triggering. Cons: if rockfall
events were as frequent as the earthquakes a rough average of the 9 candidate earthquakes
over 2000 years (759 BC1202 AD) yields an earthquake every 220 years. Such frequent
rockfall events do not seem to fit the large block sizes of blocks sampled since joints
defining these sizes probably have not been weakened so fast to re-define such large blocks
in such short time spans (otherwise the entire cliff would already crumble down by now).
DISCUSSION
81
This is also stressed by the fact that we have no age younger than 0.890.16 ka, while large
devastating earthquakes have occurred more than once in the region since (e.g. 1759 AD,
1857 AD). Considering that and the high earthquake rate epoch in the past six to eight
centuries indicated by Migowski et al. (2004) leads to suggest that cracks may not be ready
for detachment. The fact that no block was dated as younger than 0.89 ka implies that
rockfalls do not occur in such frequency, despite having candidate triggering earthquakes
over the past 800 years. OSL dating lower limit is as low as a few years (Lian and Huntley,
2001) and pit samples age results are 0.270.08 ka (0.03 m depth) and 0.710.14 ka (0.30
m depth) meaning soil from the past 800 years had accumulated on the slope, yet no rockfall
evidence was found in this study during that period. However, the latter interpretation is not
singular observations may also be explained by: a. the threshold magnitude for triggering
rockfall is higher than the magnitudes of earthquakes in that period; b. a combination of a
large earthquake event after (or during) an extremely wet period is required to trigger
rockfall, a requirement which hasnt been met over the past 800 years; c. a set of consecutive
earthquakes is required to trigger rockfall.
b. There are ages which cluster around a small number of certain earthquakes which triggered
rockfall events. Pros: assuming that there is a correlation between earthquakes and rockfall
triggering on the Ramim cliff The cons of the latter option are basically the pros of this
one: very frequent rockfall events (every 220 years) contradict two observations: very large
block sizes (up to 125 m
3
; see a above) and the lack of an OSL age younger than 0.89 ka,
regardless of large earthquakes taking place since (e.g. 1759 AD, 1857 AD). On top of that,
there is a fit between known historical earthquakes and OSL ages of soil beneath the fallen
blocks suggesting rockfalls might have been triggered by some of these earthquakes, yet
rockfalls of such large block sizes are not as frequent, since the rock-mass has to go through
a maturation phase before it can fail and produce rockfalls.
c. There is no correlation between earthquakes and rockfall triggering. Pros: rockfall may be
triggered by precipitation related events (e.g. extremely rainy seasons or severe winter
storms) or even occur spontaneously. Cons: blocks in the rockfall talus are grain (block)
supported, implying catastrophic rockfall events rather than spontaneous, sporadic block
detachment, hence there had been a triggering event for these rockfalls other than slow
weathering. Precipitation related events may trigger slope movements (Wieczorek and
Jger, 1996) but when specifically discussing rockfalls correlation is not trivial and
pointing out precipitation as the sole triggering mechanism is not a robust enough
argument. In any case, the present local climate is not of the kind that produces large
rainstorms or thousands of millimeters of annual precipitation. Figure 4-4 (Bar-Matthews
and Ayalon, 2004) shows that at least for the past 4000 years mean annual rainfall has
been similar or less than the present day (except for a high around 4500 BP). Even the
extremely rainy winters of 1968/69 and 1991/92, in which annual precipitation in northern
DISCUSSION
82
Israel was double than the annual mean, did not trigger any significant slope movements
(including rockfalls) on the Ramim cliff.

Figure 4-4. Calculated paleo-rainfall amount during the last 7000 years (derived from speleothems in the
Soreq cave). Horizontal line (around 500 mm) marks present-day mean annual rainfall at the Soreq cave
(after Bar-Matthews and Ayalon, 2004).
Concluding the above, the most probable option among the three seems to be option b clusters
of OSL ages around earthquakes, which triggered the rockfalls (these clusters are detailed later in
section 4.1.5). To suggest clustering of ages, results must be tested to rule out the possibility that
sampled ages are just a random series in time.
4.1.4 Statistical Validation of OSL Age Clustering
The 9 OSL ages (Figure 4-3) span over the past 7000 years, with a mean of 1.3 ages per 1,000
years, but the degree to which only 9 observations are representative of hundreds of blocks on
the referred talus is not clear. However, it serves to indicate a possible solution and suggests the
need for further collection of field data. The validation of OSL age clustering was obtained
performing a binomial distribution test, which gives the discrete probability distribution P(k,p,n)
of obtaining exactly k successes out of n trials. The result of each trial is true (success) or false
(failure), given the probability for success (p) or failure (1-p) in a single trial. The binomial
distribution is therefore given by
( , , ) (1 )
k n k
n
P k p n p p
k

| |
=
|
\ .
, where
!
!( )!
n
n
k k n k
| |
=
|

\ .

A success was defined when the date of a given earthquake (out of the 9 candidates in Table 4-
1) with a 50 years time window was fit with one of the OSL ages with the same error range
(50 years). Since a 50 year accuracy of earthquake dates was needed, the test was performed
only for OSL ages with historical record (last 2,800 years). The OSL ages within this range are QS-
3, QS-4, QS-6, QS-9 and QS-11. Nine earthquakes, each with a 50 year time window, span
totally over 900 years out of the given 2,800 years period. Therefore, the probability p for a
DISCUSSION
83
single random earthquake to occur within this period is p=900/2800=0.32. The number of
trials n is the number of earthquakes n=9. In five cases, a success (match between an
earthquake and an OSL age, both within a 50 years time window) is obtained, therefore the
number of successes is k=5. The fit between OSL ages and earthquakes is detailed in Figure 4-5.
Age (ka)
0 1 2 3
S
a
m
p
l
e

I
D
QS-3
QS-4
QS-6
QS-9
QS-11
5
199 BC 551 AD 1202 AD
1033 AD 502 AD

Figure 4-5. Clustering of OSL ages around historical earthquake dates. Five successes for binomial
distribution test were defined using a 50 years time window (see text for detail); vertical lines are 50
years time window around earthquake dates (red solid lines mark earthquakes selected later as rockfall
triggers; blue dashed lines mark earthquakes later disqualified as triggers see next section for detail);
black circles are OSL ages with 50 years error bars (black) and their lab reported error range(gray). The
binomial distribution is P(k, p, n)= P(5, 0.32, 9)= 0.09, i.e. there is a 9% probability to obtain such a
distribution randomly. This leads to conclude that the OSL ages distribution is significantly clustered around
historical earthquakes with 91% confidence level.
Following the above data, the binomial distribution is P(k, p, n)= P(5, 0.32, 9)= 0.09, i.e. there
is a 9% probability to obtain such a distribution. Hence, it is suggested that the OSL ages
distribution is significantly clustered around historical earthquakes (and not random), with 91%
confidence level. Ruling out two out of the five successes as rockfall triggers (502 AD and 1033
AD, dashed blue lines in figure) is discussed in the next section.
It is concluded that the observed results are not a natural random series of ages therefore it
may be suggested that the ages cluster around specific dates not just randomly adjacent.
Accepting this significance with 95% confidence may be achieved with further research
involving the collection of 2030 more block ages. It is proposed that currently obtained data
exhibits clustering, but a robust validation for it requires enlarging the statistical inventory.
DISCUSSION
84
4.1.5 Rockfall Triggering by Earthquakes
Once statistical significance of the clustering of OSL ages around earthquake dates was
established, considerations for determination of specific earthquake dates out of the 9
candidates were made as follows:
a. QS-3 and QS-4 (around 1.5 ka and 0.9 ka respectively) fit the historical earthquakes of 502
AD and 551 AD (or 659 AD with a lesser fit) and 1202 AD or 1033 AD respectively. The 551 AD
earthquake is reported at more localities along the DST than 502 AD (reported on shoreline
localities only). Since the 1202 AD earthquake accounts for severe damage in other places in
northern Israel (Marco et al., 1997; Wechsler et al., 2006), it is a better candidate trigger
than the 1033 AD.
b. QS-6, QS-9 and QS-11 cluster around 2.2 ka, which fits the 199 BC earthquake.
c. QS-5, QS-12 and maybe QS-13 (with its large error which makes it seem not enough to
qualify for age validation). These fit 2050 BC and 2100 BC earthquakes suggested by
Migowski et al. (2004). Yagoda et al. (2007), found evidence for earthquake induced slope
failure east of the Sea of Galilee with OSL ages of 4.20.3 ka and 6.00.4 ka (and 587 ka)
suggesting the area had experienced strong earthquakes (Mw 7).
d. QS-9 fits the 199 BC earthquake and the 759 BC earthquake (considering error) related to
the book of Amos and the earthquake referred by Migowski et al. (2004) as 759 BC.
e. QS-1 and QS-13 (considering error) may cluster around 6.07.0 ka, but in light of the large
error range of QS-13 it seems not enough to qualify for age validation. An earthquake event
was dated by Kagan et al. (2005) to 5.16.3 ka (constrained only by post-seismic dates, i.e.
may be older), using UTh dating of cave speleothems in the Soreq cave near Jerusalem
(Figure 4-6). This may fit QS-1 (5.98.1 ka) but uncertainty range is large. These ages might
fit the slope movement evidence dated by Yagoda et al. (2007) to 6 ka at Lake Kinneret
vicinity. All these data may point out an strong earthquake event around 6.0 ka.
DISCUSSION
85

Figure 4-6. Correlation of cave seismite ages (triangles, x-axis) with lacustrine seismite ages (diamonds, y-
axis) after Kagan et al. (2005). Diagonal line represents 1:1 correlation. Gray rectangles indicate
intersection of ages of seismites from independent records; Roman numerals indicate event title; dashed
lines denote the fit between two cave seismite ages: a. 51.052.0 ka of the earthquake numbered xiv by
Kagan et al (2005) fitting lacustrine seismite ages in other independent works (Haase-Schramm et al., 2004;
Marco, 1996), which also fit the QS-2 OSL age (55.70 7.60 ka); b. 5.16.3 ka of event xviii, which fits the
QS-1 OSL age (7.0 1.1 ka).
f. QS-2 fits around 55.7 ka (7.60) which might fit the slope movement evidence dated by
Yagoda et al. (2007) to 587 ka at Lake Kinneret vicinity. Kagan et al. (2005) date an
earthquake event to 5152 ka. They noted that their results fit independent studies dating
seismites in the Lisan lacustrine sediments to 52 ka (Begin et al., 2005; Haase-Schramm et
al., 2004; Marco et al., 1996).
Kagan et al. (2007) later added that paleoseismic evidence for earthquake related to the
522 ka event were found in 3 different sites. They suggest that this seismic event, along
with another event at 391 ka, are large earthquakes, sufficiently strong to cause cave
collapses and lake bottom brecciation in numerous and distant sites in different kinds of
sediments. They further suggest that these large events are representatives of the largest
(Begin et al., 2005) magnitude earthquakes in the Dead Sea region.
The latter supports that in spite of the fact that the Lisan seismites and the Soreq cave
speleothems are distant from the Qiryat-Shemona study area (both over 150 km away), that
earthquake may be considered as a trigger for the studied rockfall, although according to
Begin et al. (2005) the epicenter is at the Sead Sea. Thus, these 522 ka earthquakes may
pose a possible fit to the QS-2 55.77.60 ka OSL age.
DISCUSSION
86
4.1.6 Other Possible Triggering Mechanisms
Rapid snow melting and freezethaw cycles are effectively out of the question due to the mild
climate of the study area, with mean minimal February (coldest month) temperature around 6C
(IMS, 2007).
Extreme precipitation might be considered a possible rockfall trigger, given the above
considerations. However, there are several arguments that weaken this possibility: a. as stated
earlier even the extremely rainy winters of 1968/69 and 1991/92, in which annual
precipitation in northern Israel was double than the annual mean, havent triggered any
significant slope movements (including rockfalls) in the study area; b. the case with evidence for
possible heavy precipitation (5055 ka; Bartov et al., 2003) matches a seismic evidence for an
earthquake of the largest magnitude in the region around ~52 ka (Kagan et al., 2007). It is
possible that a combination of an extremely wet period followed by a large earthquake has
triggered the rockfall.
To conclude, extreme precipitation is not over-ruled as a possible rockfall trigger, perhaps in
conjunction with earthquakes, but it is beyond the scope of this study to methodically put it to
the test.
4.1.7 Conclusion Rockfall Triggering
OSL ages of soil sampled beneath fallen blocks cluster around a few dates, which are suggested to
represent rockfall events triggered by earthquakes or, in one case by an unknown trigger. See
summary in Table 4-2 and Figure 4-7.
TABLE 4-2 Rockfall Triggering Candidate Events and Evidence Type*
OSL Age Cluster Candidate Triggering Event
0.9 ka 0.05 ka (1 age) 1202 AD earthquake (his)
1.5 ka 0.05 ka (1 age) 551 AD earthquake (his)
2.2 ka 0.05 ka (2 ages) 199 BC earthquake (his)
2.7 ka 0.05 ka (1 ages)** 759 BC earthquake (ps; his: book of Amos)
4.0 ka 0.7 ka (4 ages) 2100 BC earthquakes (ps); no historic data; ; exp
6.0 ka 1.0 ka (2 ages) No ps data;
56 ka 7.6 ka (1 age)*** 5152 ka earthquake (ps); ; exp
* Evidence types are known historic earthquakes (his), paleoseismic evidence (ps) or possible extreme
precipitation (exp) events
** Dashed line separates known historic earthquakes from unreported in history and prehistoric earthquakes
*** Excluded from graph below (very large uncertainty range)
Other evidence correlated to large (Mw 7 or more) earthquakes (Kagan et al., 2007; Kagan et al., 2005;
Yagoda et al., 2007)

DISCUSSION
87
The OSL age determination aided to conclude that earthquakes are the triggering mechanism of
rockfalls originating from the Ein-El-Assad Formation and yields important results (discussed
later in text) regarding the recurrence time of rockfalls, which may serve later in the evaluation
of rockfall hazard to Qiryat-Shemona. However, not all historic earthquakes triggered rockfalls.
This may indicate that:
- A combination of a large earthquake event after (or during) an extremely wet period
is required to trigger rockfall.
- A set of consecutive earthquakes is required to trigger rockfall.

Age (ka)
0 1 2 3 4 5 6 7 8
S
a
m
p
l
e

I
D
QS-1
QS-3
QS-4
QS-5
QS-6
QS-9
QS-11
QS-12
QS-13
Yagoda et al
Kagan et al
Unknown 199
BC
551
AD
1202
AD
759
BC
2050
BC

Figure 4-7. Qiryat-Shemona OSL ages and suggested rockfall triggers. OSL age results for the past 8000 years
in black circles with error bars; ages of earthquakes determined by Yagoda et al. (2007) in gray diamonds
and by Kagan et al. (2005) in gray triangle; corresponding dates of earthquakes suggested as rockfall triggers
in gray lines and labeled at top axis (see Table 4-2). QS-2 (55.77.6 ka) excluded from graph for very large
uncertainty range.
4.1.8 Suggestion for Further Study
Dating of 9 or 10 samples is not sufficient in order to detail the history of the hundreds of rock
blocks scattered on the slopes of the Ramim cliff. A larger set of age observations is essential to
determine the validity of the correlation, which is only suggested here based on what is known
given the observations, regarding rockfall triggering by earthquakes for the Ramim cliff.
DISCUSSION
88
Considering the above, a suggestion for future work would be collecting a larger number of
samples, some located at the upper (western) part of the slope (not reachable with a tractor,
require manual excavation). Such work would help getting a clearer, more distinct picture of the
age distribution of the Ramim cliff rockfall events.
4.1.9 Seismic Topographic Amplification
The effect of topography on surface ground motion was observed at many studied earthquakes
(e.g. Celebi, 1987; Griffiths and Bollinger, 1979; Hartzell et al., 1994; Havenith et al., 2003;
Zaslavsky and Shapira, 2000; Zaslavsky et al., 2000). Close links between site effects due to
topography and damage of structures were reported by many investigators (e.g. Celebi, 1987).
Characterizing the effect of topography is commonly performed using the spectral ratio
technique at various points on a topographic feature (at the ridge top and valley bottom, and
also along the slopes relative to a reference station).
The Manara strong motion station installed on the high plateau near the top escarpment of the
Naftali Mountains ridge (some 400 m above the Ein-El-Assad Formation escarpment) was
triggered by an earthquake in the Dead Sea that occurred on 11.02.2004 (M
L
=5.2, epicentral
distance 165 km). Zaslavsky (2008) found H/V ratios to clearly exhibit a resonance peak near
2.5 Hz with amplification up to 6.0. At the NS components, amplification at 3.0 Hz was observed
with corresponding amplitude of about 4.0.
This topographic amplification of seismic shear waves supports the suggestion that earthquakes
trigger rockfalls from the Ein-El-Assad Formation escarpment, located less than 1 km east to the
Manara station. Surface ground motions induced by local earthquakes are amplified several
times within the topographic high, increasing the probability for triggering of rockfalls from the
Ein-El-Assad Formation.
Amplification is expected to be stronger at higher parts of the ridge it is therefore expected to
find rockfalls triggered by amplified ground motion at the escarpments located higher above the
Ein-El-Assad Formation outcrop. This may suggest future study of rockfalls of the Naftali
Mountains from these potential outcrops.
4.1.10 Magnitude of Triggering Earthquakes
A lower limit for earthquake magnitude required to trigger rockfall is not trivial, especially in
light of the above discussed topographic amplification of ground motion. Smaller earthquake
magnitudes occurring in the vicinity of the study area (situated on one of the Dead Sea
Transform branches) induce ground motions which, once amplified by topography, may trigger
rockfalls. A recent earthquake (15.2.2008; ML=5.2) occurred in Lebanon, some 18 km
northwest from the study area. Field survey hasnt been carried out since that earthquake, but no
evidence for rockfalls at the study area has been reported.
DISCUSSION
89
Nevertheless, it should be noted that this topographic setting has always been true regarding the
Ein-El-Assad escarpment, and still rockfalls have not been very frequent as already stated earlier.
Once again, it is suggested that strong ground acceleration is not the only term for triggering of
rockfall, but a certain stage of maturity of fracture weakening (and/or deformation) of the rock-
mass must be fulfilled in order to trigger the rockfalls. Suggested rockfall triggering earthquakes
(Table 4-2) are of estimated magnitudes M
w
6.87.5 (with maximal epicenter distances of 100
150 km respectively) according to the catalogs and paleoseismic data from which they were
extracted, thus it is concluded that M
w
7 with maximal distance from epicenter of 100 km is a
lower threshold for magnitudes of rockfall triggering earthquakes.
4.1.11 Rockfall Recurrence Interval
Recurrence times of the suggested rockfall events (Table 4-2) can be estimated by dividing 6
events over a 5.2 ka time span (last known rockfall at 1202 AD; earliest rockfall at 6.0 ka),
yielding a mean of about 850 years between consecutive rockfalls (SD=570 yrs). The ~56 ka OSL
age is excluded from these calculations since there were no OSL ages found between 6 ka and 56
ka, while for the past 6 ka there were 9 ages found. This 50 ky gap may be explained by: a. more
OSL data must be obtained; b. a very long rockfall quiescence period which does not fit the
rockfall and earthquake recurrence intervals suggested here and is therefore less probable.
Suggested triggering threshold M
W
7 falls in good agreement with Begins (2005) recurrence
intervals of earthquake magnitudes for the Kinneret-Dead Sea segment (section 1.2.2): M
W
6.5
7.0 recurrence interval is between 8003000 years.
4.2 Rockfall Trajectories and Affected Area
4.2.1 Characteristics of Block Stop Locations
Stop angles and swaths results (section 3.2.4.3) and all other cell slope angles per profile
(calculated from cells start and end coordinates) are plotted in Figure 4-8. 100% stop angles
(where 100% of the blocks stop) of all profiles (red circles in figure) have a mean of 7.7 with
SD=2.3 (minimum of 3 and maximum of 12); 50% stop angles (where 50% of blocks stop;
blue triangles) have a mean of 10 with SD=5.3 (Table 4-3). All other cell slope angles (gray
circles) vary widely between 788, among them very few are less than 10 (Figure 4-8). This
leads to conclude that a. most blocks (50% at least) keep traveling down-slope until the slope
angle decreases to 1015; b. all blocks stop where the slope angle decreases to 5.510.0
(considering standard deviation). Stop swath distances (Table 4-4) vary between 8105 m,
having a mean of 38 m with SD=24 m.
DISCUSSION
90
Profiles
0 5 10 15 20 25 30
C
e
l
l

s
l
o
p
e

a
n
g
l
e

(
d
e
g
.
)
10
20
40
50
70
80
0
30
60
90

Figure 4-8. Slope angle at 50% and 100% of blocks stop. Cell slope angles of all CRSP simulation profile cells
in gray circles; 50% stop angles in blue triangles; 100% stop angles in red circles. See Table 4-3 for means
and standard deviations.
An attempt to correlate stop angle and stop swath distance using a linear regression yielded R
2
=
0.13 (n=25 profiles) and it is therefore suggested that there is no correlation found between the
two. This leads to conclude that characterization of block stop locations should be based both on
stop angle and stop swath distance 100% block stop occurs when both conditions are met.
To conclude the characterization of rockfall stop line (regarding the discussed study area), two
possible distinctions are considered: one based on statistic properties (means and standard
deviations), the other based on extreme values (lowest slope angle and longest swath distance)
and is therefore more conservative. These are referred to in Table 4-4 below as the statistic and
the conservative approaches. Results may differ in other slopes, where the mechanical
properties of the slope and rock blocks are different.
TABLE 4-3 Slope Angle Where 50% and 100% of Blocks Stop
Percentage of blocks stop Mean Slope Angle (deg.) SD (deg.)
50% 10 5.3
100% 7.7 2.3

DISCUSSION
91
TABLE 4-4 Rockfall Block Stop Location Characterization
Approach Stop Slope Angle (deg.) Stop Swath Distance (m)
Statistic* 5 10 14 m 62 m
Conservative** 3 105 m
* Mean 7.7 stop angle with SD=2.3 and mean 38 m stop swath distance with SD=24 m.
** Lowest stop angle and longest stop swath.
4.2.1.1 Implications for Rockfall Hazard Mitigation Design
The above conclusion and angle/swath distance values may be related when considering rockfall
mitigation design. The above values mean that falling blocks would stop after covering a
distance of 1462 m in a slope angle of 510, while the conservative values mean that blocks
would stop after traveling 105 m in a 3 slope. The design of such a flat artificial topographic
surface may stop traveling blocks, but it is also not recommended since it involves an expensive
project with severe damage to the natural environment. Other mitigation acts, some better
suited for the natural environment and easier to accomplish are suggested later in this chapter
(section 4.4.2).
4.2.1.2 Mis-fit of Small Blocks Simulated Travel Distances
The longer observed than simulated travel distance of smaller blocks (section 3.2.2) may be
explained in a few ways: First, smaller blocks may be more subject to down slope creep being
more affected by water runoff and slope material movement due to their smaller size. Therefore
they travel further down after the rockfall event took place. This is re-enforced by the minimum
shadow angle results (section 3.2.4.4) which yield mean simulated shadow angle of 21.5
(SD=1.2) and mean observed shadow angle is 18.0 (SD=1.7), while typical worldwide values
are 2230 (Dorren, 2003), which leads to suggest that the longer observed than simulated
distances (resulting in shallower shadow angles) are not typical for rockfall behavior these
blocks may have traveled to their current locations by creep mechanism. Second, some of the
small blocks (<1 m) might have been misinterpreted as blocks during the analysis of 1946 and
1951 aerial photos. All blocks demonstrating such behavior were observed on aerial photos only,
since they were situated where the town is built today. As for the larger blocks demonstrating
longer observed than simulated travel distances one interpretation suggestion is also related to
the aerial photo interpretation at which 2-D block surface observations were extrapolated to 3-D
block volumes based on field observations (section 2.1.4). It is possible that blocks which
appear to have larger surface area in the aerial photos were actually discoidal (flat) and
therefore had smaller volumes and so might have been moved by slope creep down slope along
with other smaller blocks.
Another interpretation, which is suggested for all block sizes demonstrating longer observed
than simulated travel distances is based on the fact that the construction of town has created a
DISCUSSION
92
different topographical setting than the slope on which the blocks had traveled during rockfall
events at the past. This observation occurs (in most cases) for blocks on simulation profiles for
which travel distances reach town border or very close to it (see profiles 613 in Figure 4-9). A
built area involves surface leveling and the town itself has introduced a flat area where once a
slope existed. Since the simulation profiles were created from current topography, which
includes that flattening of the slope, simulated (and future rockfall) blocks no longer travel to
the distances at which they were observed before town was built. For example, the simulation
travel distances of profiles 1012 are close to town border (red dashed line), while observed
blocks (yellow rectangles) have traveled up to 300400 m further down slope. As detected from
elevation contour density (white lines), town buildings and nearby fields have created a flatter
surface which would render potential falling blocks to a stop (see blue inset in Figure 4-9). This
argument is reinforced by the conclusions about the relation between the stopping of blocks and
slope angle discussed earlier in the text (section 4.2.1). The observed blocks extracted from the
1946 aerial appear to have come to a 100% stop at a slope portion with an angle of 6. This
suffices to meet the statistic terms suggested in Table 4-4.
One should be aware of the possibility of a chicken and egg case, according to which the town
built premises developed into the already flatter topographic locations. No topographic data
dating before the construction of Qiryat-Shemona is available at that resolution (e.g. 5 m
elevation contours used in this study), but this possibility can be ruled-out by examining the few
blocks that seem to lie on an undisturbed slope further away from stop line (see green inset in
Figure 4-9). Before construction of town these blocks could travel to these locations, and after
construction, topography west to that undisturbed slope (where profiles 7 and 10 are marked) is
flatter and blocks are unable to reach it.
DISCUSSION
93

Figure 4-9. Observed blocks (from field and 19461951 aerial photos) vs. CRSP simulated travel distances.
Observed blocks in yellow rectangles (larger marks are larger blocks); simulated maximal travel distances
for each profile in orange-hazard triangles with profile numbers in yellow; town border in dashed red line; 5
m elevation contours in white (50 m contour labels); refer to discussion in text above for blue and green
insets. Most of the observed blocks located at larger distances than simulated are on profiles 613 (southern
part of town).
4.3 Hazard from Future Rockfalls
Maps detailing different aspects of rockfall hazard are presented in sections 3.2 and section 3.3.
Following is a discussion regarding their significance.
4.3.1 Rockfall Hazard Area Map (Fig. 3-21)
This map details the area subject to future rockfall hazard from the largest predicted blocks
(D=6.2 m, V=125 m
3
).
This map may be used for several aims, such as:
- Locate existing specific points or areas of interest which are presently under rockfall
hazard and may require mitigation actions.
- Check whether rockfall hazard exists for future planned areas of interest.
- Locate safe locations for any sort of future planning (e.g. buildings, parks, roads and
infrastructure).
200
100
150
250
DISCUSSION
94
4.3.2 Town Border Impact Hazard Map (Fig. 3-25)
This map details the locations where future rockfalls are predicted to impact town premises.
Since there is no simple way to simulate the moving blocks along town premises (e.g. buildings,
parkways) it is assumed that the first line of buildings/town property would suffer the rockfall
impact. Exact locations where each simulated profile impacts town premises are marked,
referring to the analysis of predicted block velocity and kinetic energy at that point (detailed also
in the engineering appendix).
This map may be used to:
- Identify locations of specific town premises currently subject to rockfall hazard.
- Use the specifically referred velocity and kinetic energy analysis to design mitigation
actions for specific localities or for future planning.
4.3.3 Frequency of Stopping Blocks and Analysis Location Map (Fig. 3-23/24)
This map details the locations of cumulative frequency of block stop distances (50%, 68%, 95%
and 100% of simulated blocks). This map demonstrates the width of the swath where most of
the falling blocks stop (section 4.2.1). The map demonstrates that most of the falling blocks are
predicted to reach similar distances as the farthest traveling blocks (except for the locations of
profiles 2425). It should be noted that for each simulation along a profile, 100 blocks were
simulated and cumulative block number data was collected, but the simulation algorithm of CRSP
v4 simulates each of these 100 blocks as a single block traveling down-slope (mutual interaction
between falling blocks is not applied). It is probable that during rockfall, some of the blocks
interact while traveling down-slope and their distribution along slope would be somewhat
different (some of the energy is dissipated in block interactions and blocks may stop earlier).
Using 100 simulated blocks contributed the ability to extract statistical aspects of the rockfall,
under a worst-case scenario. This map also includes reference points to locations where the
velocity and kinetic energy analysis was performed for each slope profile (which also refer to the
engineering appendix).
This map may be used to:
- Refer to velocity and kinetic energy analysis at specific locations at the end of each
profile (also available at the engineering appendix).
- Locate point or areas subject to larger probable hazard (where cumulative
probability of block which have stopped is lower predicted hazard is higher since a
larger number of blocks are still in motion in that location).
- Locate areas or points of interest west to Route 90 at the southern part of the study
area (beyond the southern border of town) and use local velocity and kinetic energy
analysis for present mitigation design and future planning.
DISCUSSION
95
4.3.4 Rockfall Hazard Evaluation for Qiryat-Shemona
Rockfall hazard evaluation is performed by multiplication of the probabilities of each term
required to occur simultaneously, as hereby detailed.
4.3.5 Rockfall Block Size Probability
Cumulative probabilities for selected block diameters, calculated from the p(D) regression curve
(Figure 3-6), which correspond to blocks of given sizes or smaller (P
D
) are detailed along with
the probabilities for blocks larger than same given size (1-P
D
) in Table 4-5. These may serve later
as guidelines for engineering solutions in terms of kinetic energy. Cumulative probability for
D=1.3 m (V=1 m
3
) is unavailable since no data was collected for D<1.3 m. Note that the
regression curve yields probability of 1.00 (100%) for blocks of diameter D6.0 m (V113 m
3
).

TABLE 4-5 Cumulative Probabilities of Selected Block Diameters*
Block Diameter (m)** Volume (m
3
) P
D
*** 1-P
D
***
2.7 10 0.671 0.329
4.6 50 0.891 0.109
5.8 100 0.986 0.014
6.0 113 1.000 0.000
* Probability derived from fit-curve for block diameter cumulative probability p(D)
** Probability for D=1.3 m (V=1 m
3
) is not available (no data collected for blocks <1m
3
)
*** P
D
is probability for given block size or smaller (worst-case); 1-P
D
is probability for larger than given size
4.3.6 Rockfall Recurrence Time Probability
Rockfall occurrence probability P
R
(t) is based on the OSL ages analysis, which yielded a mean
recurrence interval of 850 yrs (section 4.1.11). A 50 year period is common when discussing
engineering projects and the Israeli Building Code 413 is based on this period, therefore rockfall
occurrence probability relates to a 50 year time period noted as P
R50
.
Considering the youngest OSL age that was determined in this study (0.9 ka), the last known
rockfall was assigned the date of the 1202 AD earthquake (section 4.1.7). This means that
currently (2008) we are already in a rockfall quiescence period of 800 years. Since rockfall
recurrence is time-dependant and not random (earthquake magnitude recurrence dependant;
requires maturation of rock-mass), conditional probability analysis also used to calculate
probability for strong earthquake occurrence (Begin, 2005) should be used to account for the
800 years of rockfall quiescence period.
Assuming no event had occurred since the last rockfall until today (t time), conditional
probability for a rockfall to occur during the next t time period is introduced by calculating the
ratio between the area of the future time period (t wide) under the Gaussian density function
DISCUSSION
96
over the entire area under the Gaussian from t time onwards (Begin, 2005). For example,
conditional probability for an event during the next 50 years, after t years elapsed since the last
event is:
p { [event within 50 years] | [event after t years] } = [p(t+50) p(t)] / [1-p(t)]
p(t) and p(t+50) are calculated from the standard Z value
( ) / Z t t o =
where is the mean recurrence interval and is its standard deviation; t is the time past since the
last event (Begin, 2005).
Introducing the recurrence interval and SD found in section 4.1.11 above (to the best knowledge
of this study) leads to suggest that rockfall recurrence probabilities P
R
(t) for the next rockfall
event are P
R50
=0.065 (6.5% during 20082058) and P
R475
=0.575 (57% during 20082483).
4.3.7 Rockfall Hazard Evaluation for Given Time Period and Block Size
Rockfall hazard H
R
is calculated by multiplying the recurrence time probability P
R
for given
period t (years) by the required cumulative probability P
D
for given block diameter D (m)
( , ) ( ) * ( )
R R D
H t D P t P D A = A
The PGA value of the Israeli Standard building code 413 states minimal ground acceleration with
10% probability in a period of 50 years, which statistically equals the probability for the same
ground acceleration (or higher) at least once in 475 years (Shapira, 2002). Following these time
periods, the evaluation of rockfall hazard for 50 and 475 year periods (P
R50
, P
R475
) for selected
block sizes is presented in Table 4-6. Hazard is evaluated per given block size or smaller.
TABLE 4-6 Rockfall Hazard Evaluation for 50 and 475 Years
Block Diameter (Volume) H
R50
Probability* H
R
475 Probability*
2.7 m (10 m
3
) 0.044 (4.4%) 0.38 (38%)
4.6 m (50 m
3
) 0.058 (5.8%) 0.51 (51%)
5.8 m (100 m
3
) 0.064 (6.4%) 0.56 (57%)
6.0 m (113 m
3
) 0.065 (6.5%) 0.57 (57%)
6.2 m (125 m
3
) 0.065 (6.5%) 0.57 (57%)
* Probability for impact by given block size or smaller (worst-case estimation); calculated for 850 yr rockfall
recurrence time conditional probability for an 800 year rockfall-quiescence period (since 1202 AD earthquake)
derived from OSL ages.
DISCUSSION
97
4.4 Rockfall Impact Mitigation Design
4.4.1 Possible Design for Mitigation of Rockfall Damage
As stated above (section 4.2.1), design according to stop angles and swath distance may be
efficient, but also expensive and harmful for the local environment. Following is a brief list of
some common engineering solutions (after Berger, 2004).
- Dynamic or static energy barriers which are usually a system of cable nets 510
meters long and 34 meters high, fixed to the land by posts and cables, with some
energy dissipater system introduced (Berger, 2004). Most are capable of dissipating a
relatively low kinetic energy (E
max
=2,0005,000 kJ for the best dynamic barriers)
which is unsatisfactory compared against the block energy at town impact points (up
to 45,000 kJ) estimated in this study.
- Roadside ditches are built in width and depth depending on the height and degree of
the slope, aimed at trapping falling blocks. Large blocks (as the ones in this case)
require deep and wide ditches and this solution may be unsuitable for the study area.
- Semi-tunnels or concrete walls to trap falling blocks are made of concrete. Their cost is
proportional to block size requiring retention, and therefore they may be very
expensive under the constraints of the study area. Environmental impact should also
be considered: These measures are not-aesthetic, form barriers for animals and alter
the natural erosion pattern.
4.4.2 Mitigation by Forests
The final report (Berger, 2004) of the ROCKFOR European Project (ROCKFALL FOREST Interrelation:
Efficiency of the Protective Function of Mountain Forest against Rockfall) deals with many aspects
of rockfall mitigation. Figure 4-10 demonstrates how a single tree (in optimal location where a
falling block hits it head-on) may be of use to dissipate the blocks kinetic energy, versus blocks
that reach impact line with town border at estimated kinetic energy of 18,00045,000 kJ for
D=6.2 m blocks. In order to answer for hundreds of falling blocks, design of a forest with
appropriate tree species planted in appropriate density should be performed. From an
environmental point of view, this is the best mitigation solution.
A thorough study is required in order to implement the mitigation measurements detailed in the
above report, e.g. identifying the counterparts of the specified European tree species among the
local north-Israeli species (if they exist), or locating the areas where such a forest is the most-
effective (e.g. at the parts of the slope where block velocities are low).
DISCUSSION
98
0
500
1000
1500
2000
2500
3000
3500
4000
4500
0
1
0
2
0
3
0
4
0
5
0
6
0
7
0
8
0
9
0
1
0
0
Diameter of the tree[cm]
P
o
t

d
i
s
s
i
p
a
t
e
d

e
n
r
g
y

[
K
J
]
Abies
Pinus uncinata
Cedrus
Pinus pinaster
Picea
Larix
Pinus nigra
Pinus sylvestros -Pseudotsuga-
Acer
Fraximus
Prunus - Carpinus

Figure 4-10. Maximum energy (kJ) that could be dissipated by different tree species with a certain diameter
in cm. (after Berger, 2004).
To conclude, a stretch of wood as wide as required (depending on the results of the work needs
to be done in the designing part of such project), located appropriately (again, after a thorough
design stage) below the Ein-El-Assad Formation escarpment is the recommended mitigation
solution for rockfall hazard for Qiryat-Shemona and its vicinity. The above mentioned ROCKFOR
final report (Berger, 2004) contains a complete set of rules and tools required for such a project,
including guidance for sustaining a forest for long periods.

CONCLUSION
99
T
Chapter 5
Conclusion
he purpose of this study is to evaluate rockfall hazard for the town of Qiryat-Shemona,
which is situated on an active segment of the Dead Sea Transform fault. Hazard
evaluation is based on modeling rockfall trajectories using the analysis provided by the Colorado
Rockfall Simulation Program v4. Evaluation of rockfall hazard aims to locate areas subject to
impact and provide relevant data to enable the design of mitigation measures to protect the
towns people and property. A possible correlation to earthquakes as a triggering mechanism of
rockfalls is analyzed using OSL dating of soil samples from beneath large fallen blocks, thus
deducing the time of rockfalls.
Eight OSL ages of rockfalls for the past 3 ka cluster around dates of historical earthquakes. Earlier
rockfalls were triggered by a prehistoric event at 6 ka and by a strong earthquake around 52 ka.
It is concluded that earthquakes of large magnitudes (M
w
7) are the triggering mechanism of
rockfalls. The mean rockfall recurrence time of 850 years is in accordance with DST earthquake
recurrence interval for such magnitudes (>800 and <3,000 years). However, the fact that large
destructive earthquakes at 1759 AD, 1857 AD that occurred in the vicinity did not trigger rockfall
(none dated in this study) and despite of expected topographic amplification, suggests that not
all large earthquakes trigger rockfalls. Apparently the rock-mass has to be weakened by joints
and fractures to facilitate rockfalls. A larger dataset of OSL ages is required for a reliable temporal
association of rockfalls and earthquakes.
Field block size distribution was found to fit worldwide rockfall size inventory catalogs, i.e.
reflecting a complete rockfall block size distribution. It was used to calculate the probability of
an impact by a block of given size (R
2
=0.97).
Using the rockfall simulation program and calibrating it to the studied area, rockfall hazard
maps are compiled and analyzed. Results show that specific regions at the south-western part of
town are subjected to rockfall hazard analysis of block velocity and kinetic energy indicates that
town premises are in danger of rockfall impact, with block velocities of 1015 m/s and kinetic
energy of 18,00045,000 kJ (98% confidence) for block volume of 125 m
3
.
Rockfall recurrence is used to calculate the probability for rockfall occurrence for a given time
period. Rockfall hazard for Qiryat-Shemona for a given time and given block size (or smaller) is
defined as the product of this and the probability of occurrence of a certain block size. OSL age
analysis (850 years recurrence time and the last rockfall triggered by the 1202 AD earthquake)
lead to a 6.5% probability for the next rockfall to occur within the next 50 years, and a 57%
CONCLUSION
100
probability within the next 475 years. Evaluated rockfall hazard probability for 50 years (H
R50
) is
0.0440.065, and for 475 years (H
R475
) 0.3850.575, for block sizes of 10125 m
3
respectively.
In light of the evaluated hazard, mitigation design should be based on the location of rockfall
susceptible points of interest. The estimated velocities and kinetic energies should be used as
parameters in the design of mitigation measures. It seems that an efficient and environmentally
friendly mitigation design is forestation of the slope.
Two aspects need to be further studied: a. augmentation of the set of rockfall ages; b. design of
proper forestation for rockfall hazard mitigation.




REFERENCES
101
References
Agliardi, F. and Crosta, G.B., 2003. High resolution three-dimensional numerical modelling of
rockfalls. International Journal of Rock Mechanics and Mining Sciences, 40(4): 455-471.
Aitken, M.J., 1998. An Introduction to Optical Dating. Oxford University Press, Oxford.
Amiran, D.H.K., Arieh, E. and Turcotte, T., 1994. Earthquakes in Israel and Adjacent Areas -
Macroseismic Observations since 100 Bce. Israel Exploration Journal, 44(3-4): 260-305.
Azzoni, A., La Barbera, G. and Zaninetti, A., 1995. Analysis and Prediction of Rockfalls Using a
Mathematical Model. International Journal of Rock Mechanics and Mining Sciences and
Geomechanics Abstracts, 32: 709-724.
Bar-Matthews, M. and Ayalon, A., 2004. Speleothems as paleoclimate indicators, a case study
from Soreq cave located in the Eastern Mediterranean region, Israel. In: R.W. Battarbee, F.
Gasse and C.E. Stickley (Editors), Past Climate variability through Europe and Africa. Kluer
Academic Publishers, Dordrecht, The Netherlands, pp. 363-391.
Bartov, Y., Goldstein, S.L., Stein, M. and Enzel, Y., 2003. Catastrophic arid episodes in the
Eastern Mediterranean linked with the North Atlantic Heinrich events. Geology, 31(5):
439-442.
Becker, A. and Davenport, C.A., 2003. Rockfalls triggered by the AD 1356 Basle Earthquake.
Terra Nova, 15: 258-264.
Begin, Z.B., 2005. Destructive earthquakes in the Jordan Valley and the Dead Sea their
reoccurrence interval and the probability of their occurrence, Geol. Surv. Israel, Report
GSI/12/2005.
Begin, Z.B., Steinberg, D.M., Ichinose, G.A. and Marco, S., 2005. A 40,000 year unchanging
seismic regime in the Dead Sea rift. Geology, 33(4): 257-260.
Berger, F., 2004. Forest interrelation efficiency of the protective function of mountain forest
against rockfalls, final report, Cemagref, Grenoble.
Boyle, R., 1664. Observations made this 27th of October, 1663, about Mr. Claytons Diamond.
In: R. Boyle (Editor), Experiments and Considerations Touching Colours. Henry
Herringman, London, pp. 413423.
Bozzolo, D. and Pamini, R., 1986. Simulation of rock falls down a valley side. Acta Mechanica,
63(1): 113-130.
Celebi, M., 1987. Topographical and geological amplifications determined from strong - motion
and aftershock records of the 3 March 1985 Chile earthquake. Bulletin of the Seismological
Society of America, 77(4): 1147-1167.
Chang, W.R., 1998. The effect of surface roughness on dynamic friction between neolite and
quarry tile. Safety Science, 29: 89-105.
Crosta, G.B. and Agliardi, F., 2004. Parametric evaluation of 3D dispersion of rockfall
trajectories. Natural Hazards and Earth System Sciences, 4: 583-598.
REFERENCES
102
Dorren, L.K.A., 2003. A review of rockfall mechanics and modelling approaches. Progress in
Physical Geography, 27: 69-87.
Dussauge-Peisser, C. et al., 2002. Probabilistic approach to rock fall hazard assessment: potential
of historical data analysis. Nat. Hazards Earth Syst. Sci., 2: 15-26.
Dussauge, C., Grasso, J.R. and Helmstetter, A.S., 2003. Statistical analysis of rockfall volume
distributions: Implications for rockfall dynamics. Journal of Geophysical Research-Solid
Earth, 108(B6).
Erismann, T., 1986. Flowing, rolling, bouncing, sliding: Synopsis of basic mechanisms. Acta
Mechanica, 64(1): 101-110.
Erismann, T.H. and Abele, G., 2001. Dynamics of Rockslides and Rockfalls. Springer, 316 pp.
Evans, S.G. and Hungr, O., 1993. The assessment of rockfall hazard at the base of talus slopes.
Canadian Geotechnical Journal, 30(4): 620-636.
Freund, R. et al., 1970. Shear Along Dead-Sea Rift. Philosophical Transactions of the Royal
Society of London Series a-Mathematical and Physical Sciences, 267(1181): 107-&.
Garfunkel, Z., 1981. Internal structure of the dead-sea leaky transform (rift) in relation to plate
kinematics. Tectonophysics, 80(1-4): 81-108.
Gesner, C., 1555. e raris et admirandis herbis quae sive quod noctu luceant, sive alias ob causas,
Lunariae nominantur et obiter de alias etiam rebus, quae in tenebris lucent,
Commentariolus (A short commentary on rare and marvellous plants that are called lunar
either because they shine at night or for other reasons, and also on other things that shine
in darkness), Tiguri, Zrich.
Giani, G.P., Giacomini, A., Migliazza, M. and Segalini, A., 2004. Experimental and Theoretical
Studies to Improve Rock Fall Analysis and Protection Work Design. Rock Mechanics and
Rock Engineering, 37(5): 369-389.
Glikson, Y.A., 1966. The lacustrine Neogene in the Kefar Gil'adi area, northern Jordan Valley. Isr.
J. Earth Sci., 15: 85-100.
Griffiths, D. and Bollinger, G., 1979. The effect of Appalachian Mountain topography on seismic
waves. Bulletin of the Seismological Society of America, 69(4): 1081.
Guidoboni, E. and Comastri, A., 2005. Catalogue of earthquakes and tsunamis in the
Mediterranean area from the 11th to the 15th century. Istituto Nazionale di Geofisica e
Vulcanologia, Rome, 1037 pp.
Guidoboni, E., Comastri, A. and Traina, G., 1994. Catalogues of Ancient Earthquakes in the
Mediterranean Area up to the 10th Century. Instituto Nazionale di Geofisica, Roma, 504
pp.
Guzzetti, F., Crosta, G., Detti, R. and Agliardi, F., 2002. STONE: a computer program for the
three-dimensional simulation of rock-falls. Computers & Geosciences, 28(9): 1079-1093.
Guzzetti, F., Reichenbach, P. and Ghigi, S., 2004. Rockfall Hazard and Risk Assessment Along a
Transportation Corridor in the Nera Valley, Central Italy. Environmental Management,
34(2): 191-208.
Guzzetti, F., Reichenbach, P. and Wieczorek, G.F., 2003. Rockfall hazard and risk assessment in
the Yosemite Valley, California, USA. Natural Hazards and Earth System Sciences, 3: 491-
503.
REFERENCES
103
Haase-Schramm, A., Goldstein, S.L. and Stein, M., 2004. U-Th dating of Lake Lisan (late
Pleistocene Dead Sea) aragonite and implications for glacial East Mediterranean climate
change. Geochimica Et Cosmochimica Acta, 68(5): 985-1005.
Hartzell, S.H., Carver, D.L. and King, K.W., 1994. Initial investigation of site and topographic
effects at Robinwood Ridge , California. Bulletin of the Seismological Society of America,
84(5): 1336-1349.
Havenith, H.B., Vanini, M., Jongmans, D. and Faccioli, E., 2003. Initiation of earthquake-
induced slope failure: influence of topographical and other site specific amplification
effects. Journal of Seismology, 7(3): 397-412.
Heim, A., 1932. Bergsturz und Menschenleben. Fretz und Wasmuth, Zurich, Switzerland, 218
pp.
Heimann, A. and Ron, H., 1987. Young faults in the Hula Pull-Apart Basin, central Dead Sea
Transform. Tectonophysics, 141: 117-124.
Hoek, E. and Bray, J.W., 1981. Rock Slope Engineering. E & FN SPON, London, 358 pp.
Huntley, D.J., Godfrey-Smith, D.I. and Thewalt, M.L.W., 1985. Optical dating of sediments.
Nature, 313(5998): 105-107.
IMS, 2007. IMS website: Climate information - long term info Israel Meteorological Service.
Jibson, R.W., 1996. Use of landslides for paleoseismic analysis. Engineering Geology, 43(4):
291-323.
Jones, C.L., Higgins, J.D. and Andrew, R.D., 2000. Colorado Rockfall Simulation Program
Version 4.0 Manual. CDOT-SYMB-CGS-99-1, Colorado Department of Transportation,
Denver, CO 80222.
Kafri, U., 1989. Facies changes and karst phenomenon in the Judea Group and their implications
on the groundwater regime in the eastern Galilee, Isr. Geol. Soc. Ann. Meeting, Ramot,
Field trip Guidebook, pp. 38-46.
Kafri, U., 1991. Lithostratigraphy of Judea Group in eastern Galilee, emphasizing the Naftali
mountains. GSI/24/91, Geological Survey of Israel, Jerusalem.
Kagan, E., Stein, M., Bar-Matthews, M. and Agnon, A., 2007. A Tale of Two Cataclysmic
Earthquakes: 39 and 52 kyr BP, Dead Sea Transform, Israel; a Multi-archive Study, Eos
Trans. AGU, 88(52), Fall Meet. Suppl., Abstract T41E-04.
Kagan, E.J., Agnon, A., Bar-Matthews, M. and Ayalon, A., 2005. Dating large infrequent
earthquakes by damaged cave deposits. Geology, 33(4): 261-264.
Katz, O. and Aharonov, E., 2006. Landslides in vibrating sand box: What controls types of slope
failure and frequency magnitude relations? Earth and Planetary Science Letters, 247(3-4):
280-294.
Keefer, D.K., 1984. Landslides caused by earthquakes. Geol Soc Am Bull, 95(4): 406-421.
Keefer, D.K., 2002. Investigating Landslides Caused by Earthquakes A Historical Review.
Surveys in Geophysics, 23(6): 473-510.
Keylock, C. and Domaas, U., 1999. Evaluation of topographic models of rockfall travel distance
for use in hazard applications. Arctic Antarctic and Alpine Research, 31(3): 312-320.
REFERENCES
104
Kirkby, M.J. and Statham, I., 1975. Surface Stone Movement and Scree Formation. Journal of
Geology, 83(3): 349-362.
Kobayashi, Y., Harp, E.L. and Kagawa, T., 1990. Simulation of rockfalls triggered by earthquakes.
Rock Mechanics and Rock Engineering, 23(1): 1-20.
Lian, O.B. and Huntley, D.J., 2001. Luminescence dating. In: W.M. Last and J.P. Smol (Editors),
Tracking Environmental Change Using Lake Sediments: Basin Analysis, Coring, and
Chronological Techniques. Kluwer Academic Publishing, Dordrecht, pp. 261282.
Lian, O.B. and Roberts, R.G., 2006. Dating the Quaternary : progress in luminescence dating of
sediments. Quaternary Science Reviews, 25(19-20): 2449-2468.
Lied, K., 1977. Rockfall problems in Norway. ISMES Publication no. 90: 51-53.
Malamud, B.D., Turcotte, D.L., Guzzetti, F. and Reichenbach, P., 2004. Landslide inventories and
their statistical properties. Earth Surface Processes and Landforms, 29(6): 687-711.
Marco, S., 1996. Paleomagnetism and paleoseismology in the late Pleistocene, Dead Sea graben.
Ph.D. Thesis, Hebrew University of Jerusalem.
Marco, S. et al., 1997. 817-year-old walls offset sinistrally 2.1 m by the Dead Sea Transform,
Israel. Journal of Geodynamics, 24: 11-20.
Marco, S., Stein, M., Agnon, A. and Ron, H., 1996. Long-term earthquake clustering: A 50,000-
year paleoseismic record in the Dead Sea Graben. Journal of Geophysical Research-Solid
Earth, 101(B3): 6179-6191.
Matsuoka, N. and Sakai, H., 1999. Rockfall activity from an alpine cliff during thawing periods.
Geomorphology, 28: 309-328.
Migowski, C., Agnon, A., Bookman, R., Negendank, J.F.W. and Stein, M., 2004. Recurrence
pattern of Holocene earthquakes along the Dead Sea transform revealed by varve-counting
and radiocarbon dating of lacustrine sediments. Earth and Planetary Science Letters,
222(1): 301-314.
Pfeiffer, T.J. and Bowen, T., 1989. Computer simulation of rockfalls. Bulletin of the Association
of Engineering Geologists, 26: 135-146.
Picard, L., 1952. The geology of Kefar-Giladi. Bull. Israel Explor. Soc., "Lif" memorial(vol. B): 73-
77.
Ritchie, A.M., 1963. The evaluation of rock fall and its control. Highway Research Record(17):
13-28.
Ron, H., Shamir, G. and Eyal, M., 1997. Deformation of Margaliot block between Roum and
Margaliot faults, Isr. Geol. Soc. Ann. Meeting, Kefar Giladi, Field trip Guidebook, pp. 33-
45.
RTM, 1997. Inventaire des mouvements rocheux, Secteur de Grenoble. 38, Service de
Restauration des terrains en Montagne Grenoble, France.
Salamon, A., Hofstetter, A., Garfunkel, Z. and Ron, H., 2003. Seismotectonics of the Sinai
subplate - the eastern Mediterranean region. Geophysical Journal International, 155(1):
149-173.
Scholz, C.H., 2002. The Mechanics of Earthquakes and Faulting. Cambridge Univ. Press, New
York, 471 pp.
REFERENCES
105
Shapira, A., 2002. An updated map of peak ground accelerations for Israeli Standard 413. In:
http://www.gii.co.il/heb/Teken/report_413.htm (Editor).
Shtober-Zisu, N., 2006. Quaternary tectonic geomorphology along the Naftali Mountain front.
PhD Thesis, Tel Aviv University, Tel Aviv.
Sneh, A. and Weinberger, R., 2002. The Geology of Qiryat Shemona. GSI-36-2002, Geological
Survey of Israel, Jerusalem.
Sneh, A. and Weinberger, R., 2003. Geology of the Metulla quadrangle 1:50,000: Implications
for the stratigraphic division and the fault system around the Hula Valley. GSI/7/2003,
Geological Survey of Israel, Jerusalem.
Varnes, D.J., 1978. Slope movement types and processes. In: R.L. Schuster and R.J. Krizek
(Editors), Landslides, analysis and control. Transportation Research Board, National
Research Council, Washington, D.C., pp. 11-33.
Vidrih, R., Ribicic, M. and Suhadolc, P., 2001. Seismogeological effects on rocks during the 12
April 1998 upper Soca Territory earthquake (NW Slovenia). Tectonophysics, 330: 153-175.
Wechsler, N., Marco, S. and Katz, O., 2006. Estimating historical earthquakes parameters using
archeology and geology in Um-el-Kanatir, Dead Sea Transform. SSA annual meeting,
Seismological Research Letters, 77(2): 247.
Weinberger, R. and Sneh, A., 2004. The geology of Qiryat Shemona region, field trips guide,
Israel Geological Society annual meeting, Hagoshrim.
Wieczorek, G., Morrissey, M., Iovine, G. and Godt, J., 1998. Rock fall Hazards in the Yosemite
Valley, US Geological Survey Open File Report.
Wieczorek, G.F. and Jger, S., 1996. Triggering mechanisms and depositional rates of postglacial
slope-movement processes in the Yosemite Valley, California. Geomorphology, 15: 17-31.
Wieczorek, G.F., Nishenko, S.P. and Varnes, D.J., 1995. Analysis of rock falls in the Yosemite
Valley, California. In: J.J. Daemen and R.A. Schultz (Editors), Rock Mechanics Symposium,
35th, Reno: Nevada, University of Nevada. Balkema, Rotterdam, pp. 85-89.
Yagoda, G., Katz, O., Amit, R. and Hatzor, Y.H., 2007. Estimation of PGA values from back
analysis of a large landslide in the eastrens margins of the Kineret, Israel Geological Society
Annual Meeting, Dead Sea.
Zaslavsky, Y., 2008. Strong ground motion topographic amplification in the Manara station,
Personal discussion.
Zaslavsky, Y. and Shapira, A., 2000. Observation of topographic site effects in Israel. Israel
Journal of Earth Sciences, 49: 111.
Zaslavsky, Y., Shapira, A. and Arzi, A.A., 2000. Amplification effects from earthquakes and
ambient noise in the Dead Sea rift (Israel). Soil Dynamics and Earthquake Engineering,
20(1-4): 187-207.


ENGINEERING APPENDIX
107
Appendix A
Engineering Appendix
A.1 Block Properties
Simulated diameters were calculated from representative volume scales rounded to 1 decimal
digit. Rock density is 2,500 kg/m
3
(Ein-El-Assad Formation limestone).
TABLE A-1 Simulation Block Properties
Simulated Diameter (m) Volume Scale (m
3
)* Exact Volume (m
3
)** Weight (kg)**
1.3 1 1.2 2,875
2.7 10 10.3 25,750
4.6 50 51.0 127,400
5.8 100 102.2 255,400
6.2 125 124.8 312,000
* Representative block volume scales
** Reported by CRSP v4 (calculated from simulated diameter).
A.2 Block Travel Distances
A.2.1 Characteristics of Block Stop Locations
See section 4.2.1 - Characteristics of Block Stop Locations for detail.
TABLE A-2 Rockfall Block Stop Location Characterization
Approach Stop Slope Angle (deg.) Stop Swath Distance (m)
Statistic* 5 10 14 m 62 m
Conservative** 3 105 m
* Mean 7.7 stop angle with SD=2.3 and mean 38 m stop swath distance with SD=24 m.
** Lowest stop angle and longest stop swath.
ENGINEERING APPENDIX
108
A.3 Maps of Predicted Hazard from Future Rockfalls
A.3.1 Rockfall Hazard Area Map
This map (section 3.2.4) details the area found as subject to future rockfall hazard. It may be
used for these aims:
- Locate existing specific points or areas of interest which are presently under rockfall
hazard and may require mitigation actions.
- Check whether rockfall hazard exists for future planned areas of interest (west of the
current built area).
- Locate safe locations for any sort of future planning (e.g. buildings, parks, roads and
infrastructure).
A.3.2 Town Border Impact Hazard Map
This map (section 3.3.2) details the exact locations where future rockfalls are predicted to
impact town premises. Simulated impact locations at town border are marked, referring to the
analysis of predicted block velocity and kinetic energy at that point (section A.4). This map may
be used to:
- Identify exact locations subject to rockfall.
- Refer to velocity and kinetic energy analysis for mitigation design.
A.3.3 Frequency of Stopping Blocks and Analysis Location Map
This map (section 3.2.4.2) details the locations where specific cumulative frequencies of
simulated blocks stop (50%, 68%, 95% and 100% of simulated blocks) and includes reference
points to exact locations where the velocity and kinetic energy analysis for the entire study area
(including west to Route 90) were performed.
This map may be used to:
- Refer to velocity and kinetic energy analysis for mitigation design.
- Refer to velocity and kinetic energy analysis for future planning.
- Locate areas subject to higher probable hazard (where cumulative probability of
stopped blocks is lower predicted hazard is higher since a larger number of blocks
are still in motion in that location).
ENGINEERING APPENDIX
109
A.4 Velocity and Kinetic Energy Analysis
A.4.1 Velocity and Kinetic Energy at Town Border Impact Locations
Following are the calculated velocity and kinetic energy of a D=6.2 m (V=125 m
3
) block at
impact point with the western town border. Cumulative probability for this block size or smaller
is 100%. Referred profile numbers (locations of analysis points) are marked by orange-hazard
triangles on the Town Border Impact Hazard Map (section 3.3.2).
TABLE A-3 Velocity and Kinetic Energy at Town Border Impact Locations
Profile East ITM (X) North ITM (Y) Velocity (m/s) Kin. Energy (kJ)
8 253137 788708 11.9 30,441
9 253137 788663 14.8 24,821
10 252851 788951 9.7 29,576
11 252833 789006 13.9 40,172
12 252827 789090 11.9 17,944
13 252831 789119 12.2 44,479
14 252854 789184 11.1 28,762
* Mean velocity 12.2 m/s (SD=1.7 m/s); mean kinetic energy 30,885 kJ (SD=8,960 kJ)
A.4.2 Velocity and Kinetic Energy at Block Stop Locations
Calculated velocity and kinetic energy of a D=6.2 m (V=125 m
3
) block at stop location (95% of
total travel distance). Cumulative probability for this block size or smaller is 100%. Analysis
point locations are marked by orange triangles on the Block Travel Distance Hazard Map
(section 3.2.4.2).
ENGINEERING APPENDIX
110
TABLE A-4 Velocity and Kinetic Energy at Stop Location
Profile East ITM (X) North ITM (Y) Velocity (m/s) Kin. Energy (kJ)
5 252759 789462 5.6 6,632
6 253063 788794 8.9 12,737
7 252928 788865 4.5 4,334
8 253201 788705 10.4 17,407
9 253196 788662 11.5 21,690
10 252893 788947 8.4 11,188
11 252933 789004 7.4 10,893
12 252933 789084 7.5 11,336
13 252935 789117 7.7 11,779
14 252887 789183 6.9 9,623
15 252846 789266 6.3 6,576
16 252880 789224 6.7 7,445
17 252784 789361 6.7 9,180
18 252781 789423 7 8,086
19 252706 789599 7.3 8,740
20 252764 789719 4.6 4,479
21 252752 789815 6.2 6,541
22 252599 790074 5.1 4,423
24 253138 788424 8.6 11,972
25 252762 787997 6.7 7,598
26 253085 787451 6.2 6,492
27 253026 787273 6.8 7,619
28 253044 787104 5.7 5,400
29 253176 786601 4.9 4,196
30 252716 789505 5.7 5,464
* Mean velocity 6.9 m/s (SD=1.7 m/s); mean kinetic energy 8,873 kJ (SD=4,177 kJ)
Rockfall impact hazard to town premises
A.5 Rockfall Hazard Evaluation for Qiryat-Shemona
A.5.1 Rockfall Recurrence Time Probability
Recurrence time probability P
R
is based on OSL ages analysis of one rockfall event every 850 years.
Conditional probability is used to account for 800 years of rockfall quiescence. Calculation per
given period t is detailed in section 4.3.6.
ENGINEERING APPENDIX
111
TABLE A-5 Rockfall Recurrence Probability
Time Period t Rockfall Probability PR(t)
50 years 0.065 (6.5%)
100 years 0.130 (13%)
475 years 0.575 (57%)
A.5.2 Rockfall Block Size Probability
The probability for a block of given diameter D or smaller is
0.412 ( ) 0.262 p Ln D = +
This may also be used to determine the maximal block diameter which may be encountered
considering a given probability p
exp[( 0.262) / 0.412] D p =
This correlation should be used when considering probable rockfall hazard. Since larger blocks
pose larger hazard due to higher kinetic energy which increases potential damage, the
probabilities for different block sizes may be required when planning rockfall mitigation
solutions. Selected diameter values calculated per cumulative probabilities from the D(p)
correlation above are presented in Table A-6.
Block size probabilities calculated according to the p(D) fit-curve correspond to blocks of given
sizes or smaller (P
D
) and probabilities for larger blocks than that given size (1-P
D
).
TABLE A-6 Cumulative Probabilities of Selected Block Diameters*
Block Diameter (m)** Volume (m
3
) P
D
*** 1-P
D
***
2.7 10 0.671 0.329
4.6 50 0.891 0.109
5.8 100 0.986 0.014
6.0 113 1.000 0.000
* Probability derived from fit-curve for block diameter cumulative probability p(D)
** Probability for D=1.3 m (V=1 m
3
) is not available (no data collected for blocks <1m
3
)
*** P
D
is probability for given block size or smaller; 1-P
D
is probability for larger than given size blocks
A.5.3 Rockfall Hazard Evaluation for Given Time Period and Block Size
Rockfall hazard H
R
is calculated by multiplying the recurrence time probability P
R
for given
period t (years) by the required cumulative probability P
D
for given block diameter D (m)
( , ) ( ) * ( )
R R D
H t D P t P D A = A
ENGINEERING APPENDIX
112
TABLE A-7 Rockfall Hazard Evaluation for 50 and 475 Years
Block Diameter (Volume) H
R50
Probability* H
R475
Probability*
2.7 m (10 m
3
) 0.044 (4.4%) 0.38 (38%)
4.6 m (50 m
3
) 0.058 (5.8%) 0.51 (51%)
5.8 m (100 m
3
) 0.064 (6.4%) 0.56 (57%)
6.0 m (113 m
3
) 0.065 (6.5%) 0.57 (57%)
6.2 m (125 m
3
) 0.065 (6.5%) 0.57 (57%)
* Probability for impact by given block size or smaller; calculated for 850 yr rockfall recurrence time
conditional probability for an 800 year rockfall-quiescence period (since 1202 AD earthquake) derived from
OSL ages.
A.6 Rockfall Impact Mitigation Design
A.6.1 Possible Design for Mitigation of Rockfall Damage
Reducing rock-fall hazard by engineering the slopes to suggested stop angles and swath distance
may be efficient, but also pretty expensive and very harmful for the local environment (the study
area is also part of a national park). It is recommended that mitigation design would base on
forestation of the slope between the Ein-El-Assad Formation escarpment cliff and Qiryat-
Shemona. A comprehensive reference for rockfall mitigation using forestation is the final report
of the ROCKFOR European Project (Berger, 2004). The project website, where that report is also
available on-line, can be found here: rockfor.grenoble.cemagref.fr/texte/results.html

BLOCK CATALOG
113
Appendix B
Block Catalog
Details of the 76 blocks (V>1 m
3
) documented in the mapping area:
TABLE B-1 Block Catalog
Block ID Volume (m
3
) East ITM (X) North ITM (Y) Photo #
001 17.5 252605 789538 4166 R
002 56.3 252603 789548 4166 L
003 35.0 252610 789544 4167
004 56.3 252548 789527 4171
005 120.0 252574 789518 4172
006 47.3 252557 789501 4175
007 80.0 252571 789492 4176
008 25.0 252551 789500 4178
009 9.0 252551 789505 4179
010 16.9 252533 789503 4183
011 22.5 252626 789504 4184
012 17.0 252607 789506 4186
013 10.5 252588 789478 4188 L
014 4.3 252581 789469 4188 R
015 18.0 252603 789478 4190
016 41.3 252611 789373 4193
017 36.0 252695 789425 4206
018 5.7 252537 789523 4208
019 6.2 252537 789517 4210 L
020 15.0 252548 789534 4211
021 18.8 252529 789542 4216
036 13.2 252682 789425 10384
037 3.7 252696 789446 10385
038 9.4 252687 789492 10386
039 18.5 252738 789492 10387
BLOCK CATALOG
114
TABLE B-1 Block Catalog
Block ID Volume (m
3
) East ITM (X) North ITM (Y) Photo #
040 3.8 252750 789524 10389
041 6.7 252690 789572 10390
042 3.9 252699 789441 10393
043 6.3 252770 789405 10394
046 1.1 252764 789322 10397
048 3.6 252755 789346 10399
049 3.5 252753 789341 10400
051 2.0 252751 789291 10410
052 1.4 252749 789288 10411
054 1.4 252736 789309 10413
055 2.8 252734 789300 10414
056 1.9 252730 789309 10415
058 2.4 252700 789317 10417
059 11.6 252687 789301 10418
060 4.0 252711 789300 10419
061 1.8 252713 789273 10425
062 29.9 252554 789509 10427
063 32.4 252542 789467 10428
064 7.8 252531 789460 10430
065 1.8 252537 789455 10431
066 2.0 252536 789449 10432
067 4.6 252554 789466 10434
069 1.3 252560 789486 10435
070 2.2 252562 789483 10436
071 3.9 252562 789473 10437
074 7.4 252565 789426 10441 R

BLOCK CATALOG
115
TABLE B-1 Block Catalog
Block ID Volume (m
3
) East ITM (X) North ITM (Y) Photo #
075 1.1 252563 789436 10441 L
076 1.7 252572 789427 10442
077 6.3 252584 789414 10443
080 1.3 252621 789479 10448
081 3.2 252587 789441 10449
083 2.6 252577 789364 10451
084 2.9 252597 789343 10452
085 2.5 252652 789323 10453
086 1.5 252672 789311 10454
088 3.6 252584 789332 10460
092 8.6 252580 789353 10464
093 1.1 252586 789391 10465
095 31.0 252511 789336 10467
096 1.8 252516 789389 10468
098 1.4 252542 789538 10470
115 9.8 252372 789430 10487
118 4.7 252389 789466 10490
119 8.0 252402 789472 10494
120 3.5 252425 789433 10495
122 1.1 252418 789436 10497
123 1.4 252427 789515 10499 L
141 125.0 252761 789571 16286
142 35.4 252649 789600 16290
143 17.6 252668 789561 16292
144 2.0 253085 789025 16295






) M
w
7 ( ) triggering
mechanism ( .
,
, .
) 3-11 .(
.
- -
) 3-25 .(
) (
- . OSL ) 850
1202 (
6.5% - 50 , 57%
- 475 . - 50
0.044 0.065 - 475 0.385 0.575
) ( 10 125 .
, ,
10 15 ' \ 18,000 45,000 ' ) 98% (
125 .
, ,
. - .








- .
- ,
- 800 ) 1-2 .( , - 400
, - - - 40
. - ) .( -
1946 - 1951 0.1 150.0
, ) 1-10 .(
- ,
, .
- , : .
? .
? .
, - ?
, )
- 1946 - 1951 (
) 3-5 ( ; OSL
, ;
) CRSP v4 ( ;
-
) 3-21 ( ;
) 3-13 3-25 ( ,
- ) A .(
: . )
OSL ( ; . )
.(
-
ax
b
, b 1.17 ) 3-5 ( ,
. OSL

) 4-7 .(









-








' ' ,


:
' , ,
' , ,
' , ,







' GSI/24/08 , , "